Latent Space: The AI Engineer Podcast

Watch the full show on YouTube: https://www.youtube.com/watch?v=LL0YmIyBKSU

Full show notes and links

Legora can’t read a single trace of what its agent does for a lawyer. The data belongs to the law firm’s clients. How do you build and test an agent when you can’t see its work? Jacob Lauritzen is the CTO of Legora, one of the fastest-growing companies...

Full show notes and links

Arca head of AI Aman Singh joins Latent Space head of editorial Richard MacManus at AI Engineer New York, to discuss how Arca is building an AI-native wealth management firm. Singh explains Arca's architecture, which includes a custom harness and persi...

Full show notes and links

Watch the full show on YouTube: https://www.youtube.com/watch?v=v-d1WTqIULs

Full show notes and links

Katelyn Lesse and Angela Jiang join Latent Space to explain the engineering behind the Claude Platform: launching Opus 5.5, separating the model’s “brain” from the agent’s “hands,” scaling APIs, and building reliable managed agents. A new introduction ...

Full show notes and links

AI agents are creating more software—and that changes what a backend needs to do. Supabase co-founders Paul Copplestone and Ant Wilson join swyx to discuss running long-lived workloads alongside the database, making schemas easier for agents to underst...

Full show notes and links

Watch the full show on YouTube: https://www.youtube.com/watch?v=vhPCo-8K78U

Full show notes and links

Warm welcome to Basil Chatha, our newest cohost from the Forward Deployed side of the AI Engineering world! We’ll be ramping up a new show with him and couldn’t pass up an interview with Modal’s CEO Eric on the occasion of their monster Series D!

Two years ago, we spoke with Erik Bernhardsson about his vision for truly serverless AI infrastructure: developers should be able to run code without provisioning clusters, predicting load, or paying for idle resources. CTO Akshat then joined us to talk about Modal’s infra primitives and stack to offer those primitives and agent experience.

A lot has changed in the past 5 months. Since their $5B series C, Modal now reports it crossed $1 billion in annualized run rate, raised a $750 million Series D at a $16B valuation, and has spun up nearly three billion sandboxes. In this conversation, Erik joins Vibhu to explain the infrastructure and economics behind that growth—and why demand for compute rarely looks like the contracts used to buy it.

Watch the full conversation on YouTube →

GPU demand isn’t rectangular

A typical capacity contract gives a customer a fixed number of GPUs for a fixed period. Plot capacity against time and you get a rectangle. Real workloads behave differently: they grow, spike, and fall throughout the day. Buy enough for the peak and you pay for idle hardware; buy less and you risk missing demand.

Erik describes Modal’s bet as a kind of duration transformation: pool demand across thousands of customers, secure capacity underneath, and let each customer use what they need when they need it. The smoother the aggregate demand, the more useful that pool can be. It’s a useful companion to our conversation with SF Compute’s Evan Conrad: Evan focuses on making capacity tradeable; Erik focuses on making it usable in smaller, elastic slices.

There is a financial risk on the other side. If a provider locks in capacity at today’s prices and GPU prices later fall, it can be left holding expensive supply. Erik says he doesn’t expect an imminent collapse, but thinks the possibility matters when building the business. Jump to the GPU price discussion →

The sandbox that almost didn’t make it

Modal launched sandboxes after Erik and cofounder Akshat Bubna noticed that language models could generate code—and that someone would need a place to run it. Early adoption was modest enough that the team discussed shutting the product down. Then coding agents, background agents, and reinforcement learning created new demand. Erik says sandboxes have grown to roughly a third of Modal’s revenue, a result he didn’t predict at launch.

That growth also changes what infrastructure has to do. An agent may create an environment, run code, inspect the result, and try again many times. The latency of starting and restoring those environments becomes part of the experience. In our recent interview with Akshat, we explored this shift from developer experience to agent experience; here Erik adds the product and business story. Jump to the sandbox discussion →

Fast startup is an infrastructure feature

Elasticity only works if capacity can arrive quickly enough to matter. Modal’s serverless GPU write-up describes the machinery: a buffer of available GPUs, a custom filesystem, process checkpointing, and GPU memory snapshots. Those pieces reduce the work needed to bring another model replica online. It is a concrete example of how infrastructure decisions show up in an application’s responsiveness and cost.

We also discuss Erik’s path from Spotify’s recommendation system to Modal, how the company buys compute, alternatives to NVIDIA, forward-deployed engineers, and why code review is optional at Modal. Jump to engineering culture →

Show notes

* Modal — the AI infrastructure platform Erik founded.

* Erik’s writing — his essays on infrastructure, markets, and building companies.

* Modal’s Series D announcement — the company’s latest reported growth and sandbox milestones.

* How Modal achieved truly serverless GPUs — the startup and scaling techniques behind elastic inference.

* Modal’s resource solver — how it selects capacity across changing prices and availability.

* Scaling to one million concurrent sandboxes — an engineering account of sandbox scaling.

* Our first interview with Erik — the original serverless infrastructure thesis.

* Our interview with Akshat Bubna — infrastructure and developer experience for agents.

* The Programmable Data Center — Evan Conrad on GPU trading and physical settlement.

Timestamps

* 0:00 What Modal does

* 1:25 From Spotify to building Modal

* 4:19 Founding before the generative AI boom

* 7:43 First customers and fundraising

* 10:00 Building for inference and autoscaling

* 13:21 Where Modal gets GPU capacity

* 15:39 Renting vs. owning infrastructure

* 19:45 The risk of GPU prices falling

* 24:18 Is AI demand infinite?

* 26:12 Alternatives to NVIDIA

* 28:25 What Modal can learn from neoclouds

* 30:26 GPUs as an asset class

* 31:55 Modal’s products: functions and sandboxes

* 36:03 The case for smaller training workloads

* 41:21 How agent sandboxes are changing

* 46:48 Anthropic and sandbox security

* 49:29 Using agents to test security

* 52:51 Developer experience in the age of agents

* 57:45 What Modal’s forward-deployed engineers do

* 59:55 Modal’s approach to product management

* 1:03:29 Optional code reviews and engineering culture

* 1:05:56 Cloud, on-prem and bring-your-own-cloud

* 1:09:25 Capacity, fundraising and growth

* 1:13:07 Scaling the team

* 1:17:28 Erik’s tweets: ephemeral sandboxes

* 1:18:31 The problem with rectangular GPU contracts

* 1:21:20 Inference providers vs. model builders

* 1:23:25 Where models improve next

Transcript

Automatically transcribed from the revised edit; names and wording may contain errors.

0:00 Thanks for coming on, Eric Fern-Hartson from Modal. Could you explain the business a little bit and maybe how you got here? Yeah. So we build, the easiest way to put it is we build AI infrastructure. So we offer a compute platform and a storage platform that enables a lot of customers like Dacagon, and Suno, and Runway, and Cognition, and many others to essentially build AI applications. So what that means in practice is a lot of inference, a lot of sandboxes, sometimes training and fine-tuning, sometimes batch jobs and other things. So we have basically built a software SDK that enables customers to move really fast and build this. We manage all the capacity underneath, make it easier to deploy and scale and run these things in production. So you were on Light and Space like two years ago, a lot has happened since then. How big were you back then, like whatever numbers you can share and comparatively, how big are you now? I actually don’t remember where we were, but it’s probably like at least 50X, there’s revenue and users and users. Yeah. I use it actually for a side project that I have. Nice. Yeah. It’s like, yeah, turn 2D images into 3D models. Yeah. So yeah, just hosting a couple of models here. What do you think? Any feedback? I mean, I love it. Nice. Good. I mean, yeah, that’s why I use it over any competitors or anything. Good. So before you started Modal, you were at Spotify and also like better.com, I think.

1:30 That’s right. Yeah. Yeah. What were you doing at Spotify? I actually did a lot of different things at Spotify, but my main contribution was I built a music recommendation system. The first version I joined after school in 2008, 2009 maybe, and spent seven years at Spotify. And most of that time was building basically from scratch and music recommendation system and built up a small team and we launched a bunch of features like related artists and Discover Weekly and a bunch of other stuff. So a lot of, a lot of matrix factorization, a lot of compute intensive stuff. And arguably like a lot of what sort of later became the genesis of Modal was just building AI applications and realizing like the infrastructure f*****g sucks. Like, there’s got to be a better way to do that. So a lot of sort of, you know, the rationale for Modal was like kind of building a tool I always wanted to have throughout my years at Spotify. Yeah. So Modal was more of an offshoot of the stuff you learn at Spotify, not much from Better.com. Better.com was a little bit of a side tour. It was kind of a roller coaster, a long story about the company. We talked about it separately. But I was a CTO for five years of better and grew to a tech team from, I think, you know, two to three, four hundred. So I learned a lot about management and building teams and hiring and also learned a lot about developer productivity. I thought a lot about tools and things like that. Did a lot more like full stack stuff. But yeah, I kind of came back to my, was close to my heart, which is infrastructure and AI and machine learning after that. What was Better.com doing again? Mortgages. Okay. Yeah. How did you get into that?

3:00 I mean, we don’t have to spend too much time on that. Yeah, no, it’s fine. Like this guy, Zach Frankel, who I didn’t know at that time, ended up recruiting me to Better as a CTO and the company was 10 people. And, you know, for whatever reason, you know, the founder of Vishal, very interesting guy, very, very smart, a little crazy in mostly positive ways. We got very well along. So we, yeah, we started building it in 2015 and I spent five years growing that company and then, you know, eventually turned out that doing mortgages is pretty hard and, you know, interest rates went up very quickly after the pandemic and that cost a lot of issues at the company. But by then I had already left. That was like 2020, 2021 when you left? Yeah, something like that. Yeah. So, OK, you left before you had the idea for a model or you left and then you started? It’s hard to say. I had a lot of random ideas when I left Better and I, I actually wasn’t sure if I wanted to start a company. I was kind of just, you know, during the pandemic, I sort of almost took like a sabbatical at two kids. I mean, I still have kids, but I mean, at that time there were there were I think two and four and so I decided to hang out and took some time off and was kind of working like part-time, like hacking on certain things and wasn’t sure maybe I should do some consulting and I started building stuff. And I just got like super pumped about this idea that eventually became model. But actually both, you know, model was like, I think the 30 I built that I got, you know, that so there were like a few attempts before then that went nowhere.

4:30 But my model was like the first idea was like, this is kind of cool. Like I should do this. This is like, this could be a really cool platform. Were the other ideas like also in a similar vein like developer tools? The first thing was like some sort of web scraping thing, which went nowhere. But actually, you know, in hindsight, maybe could have gone somewhere. And was this before the chat GBT launch? Yeah. Yeah. 2021. Oh, OK. So chat GBT was late 2022, right? Then I started building, so maybe actually to rewind a little bit context, like when I worked at Spotify, I released an open source framework, Kaluigi, which is like workflow orchestration. So sort of the same categories, airflow and pre-effect index or sort of dated categories to some extent, like people don’t talk about as much. But for a while, it was like, you know, got a lot of attention. Anyway, I never turned it into business. But a couple of years later, when in 2021, when I was thinking about business ideas or things to work on, I kind of came back to workflow orchestration. And I started hacking on a bunch of ideas of, you know, running code in the, you know, like basic orchestrating code running in the cloud. And what I realized was like, you know, at the end of the day, like core puzzle pieces, not just like orchestrating the computer, it’s like executing the computer. And I was like, that’s kind of a bigger problem actually to solve. So in a way, I actually like, you know, left it on the shelf and then said, like, actually, I’m going to like go to this other thing over here and I’ll go back. I’ll get back to the workflow scheduling a little bit. But, but like, you know, I never got back to long story short and ended up building modal, which is like the focus is more like, let’s just figure out a way to make it easy to run code in the computer, in the cloud. And the focus was initially like fairly general purpose.

6:00 And it actually in a way like it’s always, we’ve kind of gone full circle more recently, but, but we focused on so many different applications. It’s kind of the idea was like, let’s just make it super easy to run code in the, in the cloud. Could be GPU code, AI, machine learning. It could also be like data stuff. And focused a lot on, you know, how do we scale it up? How do we make the developer experience awesome? Realized super early that like in order to do that, we’d have to go very deep in a stack and built on file system, built on container runtime and did a lot of, you know, solved a lot of hard engineering challenges. And then you mentioned chat, CPT, that was actually not quite the killer app. But like, if you remember, like that fall, another thing came out around the same time, which was stable diffusion and stable diffusion turned out to be a killer app for our, for what we’re building at that time. We were only four or five people at that time. So we started seeing a lot of people coming to us and wanting to run a stable diffusion model. And then we were real, that, that, that made us realize like maybe gen AI inference is actually a really good use case for us. And how are you finding it? I’m just like compulsive tweeter. I was even more like now, like I’m too distracted by building a company, but I always like tweeted a lot. I was blogged a lot. So I think I had a little bit of a following. And so, you know, in the early days, I like a lot of what drove the growth was just like me just compulsively just like tweeting about model and like what I was building and getting people excited and writing long blog posts about where the cloud is going and all these things. And I call it a bit like founder led marketing. I always like feel like you kind of have to lean into like what you’re good at

7:30 and like what I’m good at is like, I guess like just s**t posting and being terminally online. And so, you know, like work, it’s still like, you know, I think a lot of our growth is we’re just kind of good at social media and just like being, you know, online a lot. Nice. How long until you had your first customer? It took a while. Also, did you raise like early on? We didn’t raise for a while. I actually self funded it for a while, even with up until the first few employees, which in a way was like kind of nice because it may be really like frugal about money and like wanting to see like ROI. I wasn’t taking out a salary for a while for a long time. Actually, at some point it became a problem because I needed the money, but we didn’t raise for a while. And until early 2022, we raised the seed rounds. So at that point I’d already been working on it for about a year. Akshah joins the CTO and co-founder in about five years ago, about now August, maybe 2021. But then it took until like basically like late 2022 for us to see any meaningful revenue. I guess that’s why I’m proud of you. Yeah. And stable diffusion then in our case was the main thing that’s sort of, so we went from like, you know, $100 a month in like, I think like November, 2022. And the next month we did like $2,000 a month. And then like January, 2023, we did like 30,000 all of a sudden. And, you know, so finally I felt this like massive, just like anxiety thing off my chest. I’m like, maybe this is actually like we’re onto something. It feels like we’re onto something. And then 2023 we grew very fast.

9:00 And I guess we’ve been growing very fast since then basically. That 30 K a month moment, was that just like a couple of big customers or was it like very diverse? It’s just one. And I don’t know if I ever talked about like who that was and if they even know it, but I think they suspect it. It was a company in New York called Mage. They’re doing like a stable diffusion company. So at some point, you know, in the first, I think first quarter of 2023, I think there were like 80, 90% of our revenue. But then after that, you know, it started diversifying a little bit. And late 2023, we got soon as a big customer. They were a big chunk of our revenue for a while too. But then, you know, it’s over time, like we actually have really good revenue diversification now. Yeah. For the stable diffusion stuff, I remember using it for like comfy UI. I’m assuming a lot of people were using comfy UI, like running comfy UI workflows on it. Some people were, but I think the people that were using model, they tended to be more like low level stuff. Like they did like complicated stuff. What they like with models, like it’s always been pretty hackable. So they deployed their own custom workflows with adapters and custom inference and a lot of optimizations and all the stuff we had done in order to get the developer experience to be really good, turn out to be really good for also running production workloads. Cause auto scaling turns out to be actually a very similar problem. Like you want to scale up containers up and down very fast. So we just happened to be like very well positioned to capture a lot of those workloads in the early days. Did tools for auto scaling like not exist? Like why did you have to build those yourself?

10:30 In general, like my thesis is like, you know, every new sort of era of like, you know, technology kind of runs up against limitations of tools. You know, whether that’s you look back, you know, like the web or the cloud or, you know, mobile or PCs or, you know, AI or whatever. So I think, you know, it’s often the case that tools exist, but, but they’re often not like particularly good. And often you kind of need to rethink a lot of those tools in order to reap the benefits of, you know, and really turbo charge and start building for this new world. And you saw this with cloud, you know, we have to invent things like Kubernetes and Docker and things like that. And in a similar way, you know, like with, with gen AI, you could build these things on the sort of prior generation of tools. Like you can, you can build these things on top of Kubernetes and Docker and in these things, but, but, but those technologies are built for a very different type of workload. They’re built for a very sort of steady work, steady back end workloads. They have an even, you know, steady number of requests, like running a pretty low utilization because CPUs are normally pretty cheap. And so people don’t really worry so much about cost. And, but now we’re entering this new world and you have, you know, very expensive GPUs you want to manage capacity and utilization. You have very large models that you need to load over network into CPU into GPU. And, and it turns out like a lot of these tools, like, you know, like they were just not built for that. So we felt that, you know, we had to rebuild a lot of that, you know, building our own, I mentioned the file system, but we also spent a lot of time over the last few years, like building, you know, some really cool stuff around like CPU

12:00 snapshotting, GPU snapshotting and audit, we built on autoscaler and scheduler and, and, and in all kinds of capacity management tools. And we kind of feel like you have to do that because like you can’t build, you know, for this new generation of software on top of, you can’t build like a wrapper on top of the old stuff in order to do the new stuff. You kind of need to start over a little bit from scratch. So I guess like the auto scaling, like did you, did you see any like weird behaviors on the platform? Like was there anything that like broke while you guys were scaling? I mean, capacity is always from time to time been, been a challenge. Like how do you get like, you know, hundreds and thousands and tens of thousands of GPUs? So that’s, that’s always challenging. We had to go multi region very early. So I think that was like kind of a fateful decision in hindsight, like doing that and kind of just biting the bullet and building it. Like we always thought of like our capacity management as like one big pull. Like it’s like one big pull over the entire world. Like we don’t think of it as like separate clusters running in each region. And so that was like one of the like very consequential decisions we made early to kind of think about it that way. So I don’t know if there’s like been like, you know, specific issues. I mean, we’ve obviously had bugs and, you know, outages and I hate, you know, as much as I, you know, would love for a model to be a hundred percent uptime, you know, and I think recently we’ve basically hit four or five nines, but throughout the history of model, of course, we’ve had, you know, some run up against scaling issues. Yeah. The GPU market and well, actually before that. So you guys use AWS like GCP, like all these cloud providers underneath.

13:30 Like do you also buy any infrastructure? No, let’s talk a bit about like cloud capacity. Because I think it’s actually changed a lot. Like in the first few years we did, we got almost all the capacity pretty much from, from the hyperscalers spot and on demand capacity, mostly spot capacity, running in all the different regions, you know, using initially AWS, but also GCP and overtime measure. We started renting some capacity from Oracle pretty early because we wanted the eight tens and eight 100s and we got really good rates on those. And over time we added a couple more clouds, like doing reservations, but, but like still up until like actually end of last year, like most of our capacity was either spot or one year reservations. Spot means you’re just like renting it right there and then. Exactly. Yeah. And it’s also preemptible, which we had to deal with. But, but it sort of works in Agria because you can sort of just, you know, aggregate capacity from many, you know, hundreds of regions basically. So that, that, that model worked really well until we started running up again. I mean, first of all, like modal started growing like very fast as we needed more capacity, but fundamental also like the cloud market started changing a lot. So in the last year, basically now in order to get the capacity we need, we had to work with a much larger set of clouds. We actually work with almost 25 clouds at this point. And in many cases, in order to get the capacity we want, like we, we have to go into, you know, three plus year deals and you have to pay a meaningful amount of money upfront. And you’re locking in pricing on a three year term, right?

15:00 So, you know, like, and, and, and so we have to spend the entire, you know, last six to nine months, like, you know, kind of rethinking how we get capacity. We have to build a whole procurement like desk, like we have like four guys working full time basically, all they’re doing is like procuring capacity, calling Neoclouds, you know, doing deals, vetting the capacity. We’ve had to build a whole engineering team, basically 10 people just like managing like bare metal capacity from all over the world and dealing with network issues and bad drives and GPUs falling off the bus and, and all these things. We don’t own capacity. We rent all the capacity so far, um, you know, on, on, on typically this three year deals. And that’s like a big decision. Like, should we buy, should, should we, you know, or should we keep renting? I’m actually like not like mega dogmatic about it. I think renting is nice because there’s a certain amount of capital efficiency to it. If you’re buying, you know, like, I mean, first of all, there’s not this like operational complexity of it. Like you have to, you know, deal with vendors and suppliers and, and, you know, power and, and, and cables and all kinds of stuff. But there’s also, you know, financial complexity to it. Like, how do you fund it? Like, should you raise equity, you know, and, and you can raise a billion dollars and go buy hardware for it. That’s a very dilutive way to do it. Um, you can raise debt, but you know, if you raise that, you know, you typically want to have like locked in long-term revenue to, to, to pay the interest rates. That’s also not trivial. But you know, like we look at the economics, we look at, uh, the operational

16:30 complexity, we look at the risks. Uh, it’s a complex trade-off and, and we’re open to, to, to buying. We’ll, we’ll sort of, there’s other constructs too. Like we’re, it’s to get the capacity we need for, for 2027. It’s going to take a lot of work on, on, you know, going a lot of different avenues to, to lock it up. So do you have to hire a lot of like finance people to do all this stuff or not? Really? You know, I’ve had to spend a lot of time on this and we’ve had to, to, to spend a lot of time on this. And I do think over time, I mean, there, there is a lot of, you know, these are very expensive assets, uh, the hardware. And, you know, I think, I mean, it’s kind of staggering. You starting to see like AI investments starting to become like several percentage points of, of, of GDP of the U S and, you know, starting to see like interest rates in, in America, shooting up in the last few weeks. And I mean, I like, at this stage, like, you know, and you go and look at the Neoclouds, like these are very, very sophisticated financial engineering companies. And so yeah, we, we look at that and, you know, we look at all the options and, and it’s possible we might have to do similar things ourselves. I wouldn’t want to like, I would rather focus on being, you know, like a very lean software company, like doing the things we’re good at. I don’t think it’s good in general for software companies to own hard assets. You know, in the same way as like, should you own your office? Like I don’t think that’s necessarily the right solution for most companies. But of course, like in order to get the capacity you want, like, you know, everything is on the table. Yeah. And I guess the supply is going to be constrained for a while.

18:00 So you kind of have to do that. Like, do you, do you agree with that? Or do you think this is just like a moment in time? I don’t necessarily agree with that. I mean, like Nvidia is delivering, you know, several hundred thousand black walls, you know, a month, right? And, and I don’t want to say like the market is like 100% efficient, but like, you know, in general, like if you’re willing to pay a high price, you can get GPU capacity, right? Like if you’re willing to pay more than everyone else, and, and of course, Anthropic and OpenAI is buying up as much as they can. And, but fundamentally, like there’s constantly supply coming on the market. And so I don’t necessarily think it’s like a fully cornered resource. Prices are going up. And I do think it’s worth kind of reflecting on that. It’s like the normal state of hardware is like, it’s like out of like gravity. Like eventually, hardware prices always go down. So we’re a little bit in like a weird sort of period of time where like hardware prices are going up. And, you know, it’s hard to say, you know, how long this is going to last, you know, could be another six months, could be another 36 months. Like it’s really hard to know how long. But I fundamentally don’t know if like it’s true to say, like, you know, you need to buy capacity in order to me, that’s a little bit to sort of almost like the great toilet paper shortage of the pandemic or whatever. Right. Like, you know, at least this like, you know, run on the bank type situations. And, but that being said, like, you know, capacity is going to be tight for a while. And, you know, prices are going up. And, you know, luckily, I think, you know, it’s not even necessarily bad for us. If prices go up, because, because, you know, scarcity also means like we can

19:30 charge more from our customers. And, you know, and the fact that we have a lot of reservations locked in at a rate means like, to some extent, we’re effectively like long GPU prices already. So I don’t necessarily worry too much about prices going up. I do worry more about actually prices collapsing. I don’t think that’s going to happen anytime soon. But I think that would be a quite bad thing for a lot of the players in the market. Why would that be bad? I think it would be bad for if, for anyone who’s locked in a lot of supply. Yeah. Because they paid way too much and they paid too much, you know, so there’s a risk of buying at the top, right? And again, like, I don’t worry about that much. Like, I don’t think it’s going to be like, you know, it’s going to happen anytime soon. But like, you know, like I said, like hardware prices in the very long run, they tend to go down. And so it’s, it’s worth, you know, reflecting on that, right? Right. Like right now, I think, I think your, your capacity is constrained and, and that leads to sort of, you know, in economic terms, there’s a scarcity rent, right? Like who captured the scarcity rent? I think the primarily goes to two places, mostly Nvidia, but of course, but also to some extent, Neoclouds, like Neoclouds and in general, like, you know, anyone owning capacity has better economics. Like that’s undeniably true today with capacity being constrained. Long term, you know, in the very long run, I don’t think that’s necessarily entirely true, right? Like there’s always going to be a little bit of a premium you can get from owning GPUs, but is it worth the risk? You know, is it worth the risk of like holding this like very expensive

21:00 assets on your balance sheet? I’m not so sure, but you know, we’re going to see, like it’s going to take a few years to get through this. And, and, and in the long run, you know, we might own some, we might rent some, maybe a combination, maybe do some other crazy stuff, joint mentors, who knows? Maybe some complex SPDs, whatever. Um, everything is on the table. How do you compete with open AI and enthropping and buying GPUs capacity? I’m assuming they’d want to like buy up literally everything that’s available. They are. And, and I think what you’re seeing is, um, you know, a lot of that, the, the price is going up. Is it a reflection of that? That being said, you know, NVIDIA is shipping a lot of stuff. I think also like to some extent, like NVIDIA is like kind of prefer a little bit of a diversification. Like they don’t want like, you know, small amount of large buyers in much, in the same way as you see, you know, hyperscalers getting a little bit of a constrained allocation. Uh, I think, you know, also NVIDIA prefers to see, you know, a lot of the neoclouds and a lot of alternative clouds, like getting more allocation. And I think a little bit, what we see in the market is also this is more speculation, but I think we see, you know, smaller clouds that are a little bit more sort of risky to work with being more of an opportunity for us. Like where, you know, maybe NVIDIA, sorry, enthropping and opening AI, my feeling is like they prefer to buy from like the, the, the try and through, uh, part of it is also because they had to do a lot of training runs. So they need to have like very large blocks, but there’s plenty of capacity out there, you know, you can get 32 nodes in Armenia.

22:30 You can get 128 nodes in Indonesia. Like, you know, like we love that kind of s**t. Like, you know, so we can go out there and like get a lot of, you know, smaller capacity all over the world. Pretty good prices. Is it pretty much just like if someone offers you like a certain number of GPUs, like that’s all you care about. Is there nothing like, oh, you care about like how reliable it is, you know, is it just some like janky warehouse that someone has these in? I mean, sometimes it’s kind of a janky warehouse. It feels like, you know, but, but like, we do have like a spec sheet, basically. Like when we talk to vendors, the first thing we do is we send them a spec sheet. We say, we need this number of IOPS. We need, you know, like the CPU needs to be at least, at least, you know, have, have this many, you know, whatever the pass mark scores or whatever it is. We need to see like, you know, minimum, you know, North, North, South, bandwidth, egress and ingress. So we do have a list of requirements. Often we try to do like an acceptance testing period where we get like one node and, and runs and tests over time, you know, like to some extent, it also involves a little bit of a leap of faith. You know, sometimes we sign very big commitments, you know, 10s or 100s millions of dollars and we start working with the vendor that we still don’t fully know. And, you know, over time, like, you know, we’ll see, you know, and, and, and I think over, like, in a little bit of, like, there’s a little bit of sort of just, you know, our own sort of, like, you know, how like semi analysis has like a cluster max, I kind of feel like we almost like have our own, like modal max score internally. Like, I know which vendors I like more than others, you know, like, and, and,

24:00 and we sort of prefer to do additional deals with the vendors that we’ve seen, you know, very good reliability, not just that also, like it’s not just reliability, it’s also, are they very responsive when, when we have, you know, support issues, do the fixed issues we ask for to turn around time. If things go down, you know, how quickly do they fix it? Like things like that. Do you agree that the demand for intelligence is basically infinite? And if so, like, why wouldn’t you just buy up as much as you, as much capacity as possible? And then, like, if you over buy, you can just like sell it to somebody else. I think it’s very easy to say when there’s like a scarce resource, like, let’s just buy as much as you can. But I think, you know, first of all, like, if we bought it today, we couldn’t utilize all of it. Like we could use utilize a lot of it, right? But like, But you can sell it off to somebody else who wants to use it, right? I think the challenge is like, if there was like something in the market that happens, there are a lot of correlations that happen at the same time. If, you know, prices going down and utilization going down and the market for reselling capacity goes, vanishes at the kind of the same time. I’m sitting here and like, I’m extremely aggressive, like buying capacity every day and like raising money to buy, you know, so overall, I’m, I’m very bullish, but like, I do want to point out and like, you know, like there are tail risks here. And when those tail risks happen, like it could be get very bad. Part of it is like, maybe I’m, you know, I’m just old and I’ve seen this in the past, like I actually had my first job in 1999. So I saw the dot com crisis and I’ve seen the great financial crisis and things like that. And, you know, like, I want to build a company that, that is around for like 20

25:30 to 30, 50 more years, right? Like, and so, you know, I think we can certainly go out and like buying an insane amount of capacity right now and trying to just corner the market. But, you know, do you know about the Hunt Brothers? It’s like a legendary story in trading. They try to corner the silver market. It fails spectacularly. You know, so it’s a cornering the market. I think in general, like it can be not, not a good idea. And, but, but we are in general very, very aggressive, but we’re ramping up, you know, like at an aggressive plan that basically matches our demand growth and matches where we want to try to get to at the end of next year in terms of run runway or run rate. Are you only buying up NVIDIA stuff or also like, you know, maybe a capacity from other types of companies like training them and we’re looking at it. I think the alternative accelerators are starting to get very interesting. It’s, you know, ironically, like every, every year in the last five, maybe even 10 years I’ve been saying, you know, like next year, it’s going to be the year of like non NVIDIA accelerators. And then I’m always wrong. But, but I do think 2027 might be the year of non NVIDIA accelerators. Finally, so we’ll see if I’m wrong this year too. But so I, you know, I am these are starting to get like really good scores on certain things. I think, you know, TP use are obviously used very, you know, widely internally at Google. I seem to work really well. I don’t know much about training. I don’t feel like I see quite as, but, you know, the team is very good. I know they’re working on a lot of the software stuff.

27:00 You know, hardware is hard in a certain way. Like software is like easier. And that makes me feel like, you know, if we just fix the sort of software layer for these accelerators, like, I think there’s going to be a lot of demand for them. And I think fundamentally, you know, with NVIDIA being such a bottleneck, like, we’ll probably see, you know, more alternative accelerators. That being said, you know, there’s bottlenecks further down the pipe. I mean, you know, there’s like TSMC and ASML, maybe like a deeper bottleneck to all of this anyway. But, you know, long term, I think monopolies don’t always last. And so I think you’d be healthy for the market to have more alternative. As much as I love NVIDIA, like, to make that clear, like, I mean, they’re obviously an amazing company and within the amazing product. I do expect to see a little bit more diversity in the market as well. Also, things, you know, like Cerebras and Grock and these things. Yeah. What are your thoughts on like some of those providers? I think they’re pretty exciting. And, you know, and you see them doing really well on certain workloads. And, you know, there’s some interesting, you know, people doing these like hybrid, you know, disaggregated inference where they run, you know, NVIDIA for the pre-fill and Cerebras for the decode and things like that. Oh, I haven’t heard of that actually. Yeah. And, you know, Grock too, like, because a lot of these things, you know, certain things are compute bound, certain things are memory bound. So you can sort of mix and match. And I think that’s quite exciting. It starts to get very complex when you’re deploying these things in practice. But I do think there’s a lot of, there’s a lot of, you know, opportunity and these sort of, you know, more complex hardware setups. Are there any learnings that you get from some of these Neoclouds, like, I don’t know, like Nebius or whatever, and like the way that they’re doing things versus

28:30 like what you’re doing? I think one of the things that, you know, is very interesting with Neoclouds is like, how can they expand so quickly? I think actually the most crazy example is like N scale, which went from like nowhere to like, you know, having like, you know, whatever, hundreds of billions of dollars in the pipeline, despite barely having any capacity. How can Neoclouds grow so fast? I mean, there’s like an, you know, and basically the answer is like, you know, and this might be obvious, but like what they do is like, they go out to like, you know, enthropic or open the eyes and some big buyer or whatever. And basically say like, Hey, do you want to buy like five years of chips? You know, sign this contract, we give you a good rate. And then they take that contract and then they go to, you know, Wall Street and say, look, we got this like, you know, big buyer, you know, they committed to $10 billion worth of chips. We just need to monitor it right now. And then they go on to raise the money. And then they like, you know, go to like Super Micro and DAL or whatever in the video and like get the hardware and then they install it. And then, you know, now they have like, you know, delivered to the customer. Right. So you have this like very nice sort of financial construct where you can basically sort of pre-sell things that you haven’t even bought. And that means, you know, you can scale up extremely fast. And so that’s what we’ve seen with these Neoclouds is, is, is they can scale very, very well. I think modal because we’ve always had a little bit more of a PLG focus and, and not the sort of, you know, ability to lock in long-term contracts in the same way, it’s a little bit more challenging from a financial perspective to scale that fast, even though I think, you know, in the long run, like we’ll get to the

30:00 same point. But I think that’s been kind of a fascinating thing to see with these Neoclouds and especially like CoreWeave and those big ones. They’re very, very savvy with, you know, how they raise money and how to, you know, think about GPU as collateral for loans and all these things. What’s the relevance of, like, I forget who it was, but like some like financial company was like, oh, we’re going to make, oh, I think it was Nvidia. Actually, they were like, we are going to turn GPUs into an asset class. Is that just because like they need to raise money and they’re like trying to figure out where to get the money from? Like, what’s the relevance of that? I mean, there’s a bunch of startups. I don’t know if you were thinking about that. There’s like Silicon Data and computer exchange and like Orn and, and, and a few other ones that are like, basically like trying to get to like, you know, creating price indices and indexes and trading derivatives and things like that. I think it’s interesting. Like, I think with compute, like it does, there are a number of challenges with compute as an asset class. Like it’s definitely not as homogenous as other asset classes. You have different network and different locations and, and, and different disks and CPUs and all kinds of stuff and reliability. I mean, that being said, like you can say the same thing about oil and like we kind of created like, you know, relatively commoditized, you know, tradeable assets out of oil. It remains to be seen. Like I do think, you know, at the scale where we are with the sort of AI, like data center and GPU deployments, like we do need to be a little bit smarter with like risk hedging and, and, and how we share risks and, and, and, you know,

31:30 having more like sort of insurance and, and things like that. So I, I, I’m very happy that people are thinking a lot about this. That being said, like I’m, I’m not entirely sure if we’re going to see, you know, GPUs becoming sort of, you know, whatever it is, sort of Henry hub, like, you know, Brent crude type, like, you know, super commoditized, like tradeable asset. Yeah. I want to shift the conversation to modal sandboxes. Yeah. I guess like before we do that, what are the different products that modal has? It’s, I guess like the main product obviously is like the, the GPUs. Then you have modal sandboxes. Is there anything else? I think of the like the two big products that we have as basically number one is modal functions and number two is modal sandboxes. So, so with modal functions, we launched that, you know, back in 2021. It was the idea of like serverless functions that are very easy to deploy and very easy to scale and we managed all the capacity. That turned out to be really, really good for inference and it turned out to be really good for managing GPUs and running on things like that. But we actually never had like, you know, dedicated like inference product or like a GPU product per se. Like we never thought of modal as like a neocloud and I think there’s like a risk of like getting sucked in that direction and becoming a neocloud, like just selling, you know, sort of what I think was like rectangular, like here’s like a three year. You get this many GPUs. Models has always been very good at like you deploy something you can kind of scale it up. You have this like very cyclical or like bursty stuff. Like we manage all the capacity. And the beautiful thing is like, because we have many, many thousands of customers,

33:00 we can sort of aggregate that capacity and get very smooth workloads, you know, very smooth aggregate demand, which we can manage very, very efficiently. So modal functions was the core. We launched that in 2021. Like I mentioned, we started seeing product market fit in 2023. And now, you know, it’s about two thirds of our avenue and empowering a lot of, you know, in practice, it’s mostly inference, but you can use it for all kinds of other stuff. Modal sandboxes, we launched that in 2023. And it’s actually kind of, you know, took a while to see product market fit similar to functions. Why did we launch it? We realized a lot of people we talked to were using LLM to output code. This is like me and Akshar talking over lunch one day. We’re like, maybe they need a place to run that code. And we realized like, actually, we kind of have like a lot of the infrastructure already to just take code and run it. Because we’ve already built the primitives for like managing containers and isolation. We already had, you know, G visors and isolation primitive and things like that. So we get kind of excited about that idea. And Akshar wrote a prototype and launched it in basically one day or two days or something like that. Zero. Like we saw like very little like growth. I don’t want to like trivialize it. Because we did see some growth. Like there were like some sort of, you know, And it was 2023. So yeah, it was 2023. And like, you know, like some people picked it up. Like we had some early users who kind of enjoyed it for certain things. But like overall, like it was like a pretty insignificant source of revenue into going into the summer of 2023. So two years later, we even discussed like, do we even need this product?

34:30 Should we shut it down? Then it just started exploding. So 2023. And I think what really happened 2023 was like, LMS got really good at generating code. In 2023. 2023. I mean, it was like around that time. Like I feel like that was like when they went from like being able to like refactor functions and like write snippets, so like generating like full apps. Right. And so we started seeing a lot of growth coming from like vibe coding, you know, companies like Loveable and other ones. And then, you know, companies like ramps started using model for background agents. And then we started seeing tremendous growth coming from a number of Neolabs doing things like reinforcement learning on model sandboxes. So sandboxes is now about a third of our revenue. You know, we considered it like basically like our second product. And to me, it’s like really exciting because like even back in the early days of model, I always knew that model, you know, is about like creating a whole toolkit of products. Like we always wanted to be almost like a cloud, you know, you have the storage and the compute and a bunch of different ways to work with it. We always wanted to be a multi-product vendor, but that wasn’t necessarily true in practice until like now. We’re like, I think now it’s like kind of clear. We have functions, we have sandboxes. We’re working a bunch of other stuff too. We’re working on training, working on batch, we’re working a bunch of other stuff. So to me, that sort of evolution into becoming like a multi-product company, I think it’s been amazing because that was always so core to the vision. Even in the very early decks, I had a bunch of slides on this. But it wasn’t necessarily true to the external world until recently.

36:00 Do you have anything in the works that you think is going to be like that next product? Any predictions maybe about the market? Anything like that? I mean, I’m very bullish on training. I think training is like kind of an underserved market. Obviously, there’s like a lot of focus on like very large training runs. But we actually see a lot of demand, the sort of smaller and mid-sized training segment. So I’ll be talking a little bit about that later this week. We’re organizing a conference. Batch is I think like a really interesting market. I think we see a lot of demand for sort of, you know, customers that need a... You know, we had a customer the other day that needed to transcribe, you know, 5,000 years of audio. They had like, you know, millions and millions of audio files. And we had a lot of capacity. So we let them do that, you know, and took a few days. I think there’s some really interesting use cases there. Very bullish on like low latency stuff. I think there’s going to be a lot of, you know, once we sort of get the audio, like the speech-to-speech models working, like the voice models. I sort of expect that to be very interesting market. Then of course, like all this like agente stuff, right? And in a way, like I have like mixed feelings about like the word agente, because I feel like in certain ways, like, I think all software will become agente. So it’s almost like to me, like, what does it mean to be agente? Like I think all software in a certain way, like will become like AI powered over time. But like clearly there’s like a set of like building blocks of like, you know, doing more sort of LLM powered software that seems to be sort of, you know, common across all the different use cases.

37:30 And we’re like, you know, the way I think about models, like we’re in the business of building building blocks, you know, we want to give you like amazing, generalizable, battle-tested, you know, general purpose, composable building blocks. And I think there’s some really interesting stuff also, like, you know, with the agent use cases. That’s interesting. You bring up training because I think in another podcast, maybe like a year ago, maybe like two years ago, you said training you were not focused on like specifically, like what changed in the market? I mean, we’ve been building a training product for multiple years on and off. So I wouldn’t say we’re like, we’ve never been focused on it. But it’s like, it’s like one of those things we’re like, I mean, now we’re like 150 engineers, but, you know, a year or two ago, we were a much smaller number of engineers. And so, but just virtue of, you know, being smaller, we had to focus a little bit more. And so inference became like our sole focus and was our sole focus for many years. But we’re bigger now. Like we, we, we do have opportunities to also place bets in other markets. I’m also really interested in, you know, kind of continuing the list of things. Post training, reinforcement learning. I think there’s some really exciting stuff around that. I think there’s also many other like stuff around like storage and building data warehousing and feature stores and real time streaming pipelines and all kinds of like, I don’t know. I think the longterm view of, of modal is like, basically like, you want to build some cool AI s**t, like, you know, we got you covered, like, you know, we got all the building blocks, like comp tasks and like, we’ll help you, you know, take all these building blocks and compose them.

39:00 And that’s the beauty of software is like you, you can, you know, build almost anything if you have the right building blocks. So that, that’s very much the sort of company and the product that want to build. When you say training, are you focused on more of like the bigger enterprises doing huge training runs? No, no, they’re going to go and like, you know, do these like, you know, big, hyper-scaler or rather Neo, Neo cloud mega deals and get, you know, 100,000 GPUs or whatever. But, but, but there’s like a little bit of an underserved market of like, in my opinion, you know, like researchers iterating or like people training smaller models, there’s, you know, Texas beach and speeches, text models, computational biotech is actually a market that we see a lot of demand from. We have a lot of customers in that space training relatively, you know, modest models. You can train on 64 GPUs or whatever recommendation system, a lot of sort of traditional machine learning. There’s a lot of things like you don’t need a, you know, 100,000 GPUs. That market, I think is like, we’re not going to like focus on that for a while. But, but I think there’s a lot of stuff in the, you know, slightly smaller markets that that are relatively underserved. It’s very hard to set up training clusters. Yeah. Have you noticed like since, I don’t know, like GLM 5.2 or like, maybe like, can we 2.5 that, you know, a lot of people are using those and like fine tuning them? If anything, I’ve actually seen the other direction. I feel like fine tuning the value prop a couple of years ago was like, yeah, you know, open source models, they’re not quite as good. But, you know, if you take, you know, a small model and you fine tune it on your specific

40:30 domain, you can actually reach the frontier in terms of, you know, quality and cost. Today, I would say like the open source models are so good that like you can basically just like swap them out and just run, you know, instead of, you know, the frontier ones, you can run, you know, Kimi K3 or whatever, DeepSeq 4.1 or GLM 5.3. And they’re like basically almost as good as like the frontier model at a much lower cost. So we still see a lot of demand of for post-training. And I do think it’s a really interesting space and we’re working on some products in that space, you know, both sort of SFT and but also reinforcement learning and things like that. I would also say, you know, fine tuning is no longer the thing that you kind of have to do. Like it’s very optional. Like you can open source models like work really well just off the shelf too. Back to sandboxes actually. I forgot to ask you. So like what did you have to update about the sandboxes that you guys were offering that like existing sandboxes did not did not allow for? Not a whole lot. I think the thing we’ve always been very, very good at, I mean, we’re infrageeks, right? Like we’re very good at like, you know, scaling and reliability and doing a lot of stuff. And I think that’s why that explains to a large extent like why we have, you know, in my opinion, I don’t have numbers, but like I think we have by far the largest market share in this space. Like we’re just very good at running, you know, hundreds of thousands or even millions of sandboxes. Yeah. I will say we’ll have to, we’ve had to rewrite our systems, you know, a bunch of times in order to handle that scale.

42:00 But we were always like very good at it. There are other things like over time. I mean, the sandbox space in general, like, you know, like it’s like not the most like differentiated space, but you know, we’re working on some really cool stuff. Like over time, like we’ve had to add support for VMs, like virtual machines. So we previously bought our sandbox product on GVisor, but now we offer full virtualized machines as well. Turns out a lot of for certain use cases, you kind of need full Linux compatibility. There are other things we’ve had to build, you know, better primitives for like more interactive, you know, piping data back and forth between sandboxes, working with file system. We’ve had to spend a lot of time on snapshotting. I think a lot of the things that over in the long run will differentiate will be like, you know, really cool things like, you know, forking sandboxes, snapshotting, resuming snaps, like live migrations. In general, like I think sandboxes, the novelty of sandboxes is always, you know, this sort of ephemerality, like you can spin off sandboxes for a few seconds. And that’s something I like, that’s a string I want to pull more in. I think there’s a lot of really cool stuff you can do, you know, kind of decoupling the agent and the harness decoupling it from sandboxes, having agents control many sandboxes, forking sandboxes, you know, snapshotting things like, you know, like running basically like very short lived, maybe one per command, like things like that. So I think there’s a lot of cool stuff there. We’ve already spent a lot of time doing that, you know, around snapshots. But I think there’s a lot of cool stuff to be built there as well. You brought up snapshotting a couple of times.

43:30 Like what do you see as like the use case for snapshotting? Like what would that give us that we don’t get right now? I think the big thing for me is like today a lot of sandbox workloads are kind of like lift and shift. Like people think a lot about sandboxes almost like, you know, I have my dev box, now I’m going to shove it in the cloud, I’m going to have an agent control it, maybe I’ll run the harness inside the sandbox. I don’t know if like long term that’s like the right way to think about, you know, on like compute for agents. I think a much better model would be you’re going to have this like very long running agents, which almost like to me are like kind of durable execution workflows. Like they’re going to run, you know, 24 seven, you know, every once in a while they’re going to run commands and, and you know, having the ability to like spin up a sandbox in, you know, 100 millisecond, run a few commands and then like shut it down and like snapshot it. I think the faster you can get that like, you know, wake up, you know, go to sleep cycle, the more you can sort of create sandboxes that only run for like a very brief period and that will make compute much more efficient. And it will also enable you to do some really cool things, you know, like forking and moving things around and, and, and, you know, having like imagine, you know, like Claude code instead of, you know, having sort of a very linear flow, like sequential flow, like kind of branching and, you know, running hundreds of agents at the same time. I think a lot of like CI is like another like illustrative example, like for most of the, the history of CI, you know, talk about like GitHub actions and things like that.

45:00 Like you had a very sequential flow, like, you know, you slowly install a bunch of stuff, then you run the test suite and then, you know, generates some reports or whatever. And it’s like, you can break it up into shards, but it’s kind of annoying. Like what I would love to do is like take your entire test suite and just run every single test in its own sandbox, right? Like, and just paralyze the s**t out of that around like a thousand sandboxes. Why can’t that be done? I mean, the problem is like there’s too much, there’s too much overhead around like, you know, starting and stopping sandboxes. So I think the more like you create like, you know, frictionless, like starting and stopping, the more you can enter a world where like you can do this like extremely bursty, like large scale, like run as much as you think you want in parallel. So, you know, I’m an infrageek, like I said, like I get very excited about these things. You know, once we figure out all these primitives and the right building blocks, I think it’s going to unlock a lot of really cool stuff that, you know, we previously could not do with software. Why hasn’t that already been built? I would assume even previously people would want to parallelize as much of their CI as possible. You can’t, but the overhead, it’s like Amdahl’s law or whatever. Right? Like the overhead of parallelization right now is very high because like right now, like you can get up actions, for instance, like we parallelize in the modal test suite, right? But like what that basically means is like now if you take your test suite and you split it up in two, like now each of those shards both need to do all the installs and the test setup. And that takes a long time. Like it can take minutes, but let’s say you can do the test set up once

46:30 and then you snapshot the container and now you clone it and now you can run the test suites independently. I think a lot of the benefits cannot be, you can’t materialize those benefits until you get the friction and the overhead down, the transaction cost. I think you guys have a partnership with Anthropic and they’re using sandboxes for something, like modal sandboxes. Yeah, I mean the managed agents with Anthropic, this is sort of bring your own sandbox integration. So you can use modal for those, yeah. How did that come out? Like come to be? I think Anthropic is like trying to create like a thriving ecosystem and community and so I think they make money from inference, right? Okay. I think it’s very much in their interest to promote, for everything else, as much of an ecosystem as possible and they want to have the largest possible surface area as long as you use Anthropic for inference. So I think it makes like very rational sense for them to partner with us as a sandbox provider. Also, I feel like we have to bring up the hugging face incident. Are you okay talking about that? Yeah, yeah, I mean it’s all there. Yeah, yeah. So yeah, could you just briefly describe like what happened and I think they were using the modal sandbox but it wasn’t your, you know, it wasn’t like a fault of the sandbox. Yeah, exactly. I forget all the details of the exploit but I mean, you know, I think what you’re referring to is the fact that like modal was involved. Yeah. And so, you know, just to comment on that, like basically the agents running inside,

48:00 you know, that exploited hugging face, they discovered a vulnerability in an app that was deployed on modal and they used that to sort of, it’s almost like a springboard for getting access, getting full internet access is my understanding. I don’t remember the full details. To be clear, this is an app running a modal that wa…

Anekha Sokhal, CEO of Moshi, joins swyx to discuss building a recommendation engine for commodity traders, turning market data into useful signals, and designing AI that fits users’ workflows.

Watch: YouTube · Moshi · Anekha

0:00 AI Engineer New York and “you are the team”

0:54 Meet Anekha

2:09 From mathematics to energy derivatives

3:46 Fulbright and becoming a founder

5:45 What Moshi does

6:10 Market data

9:03 Bloomberg and market workflows

10:03 Bloomberg, Claude and Netflix

12:03 From dashboards to workflows

15:03 Trusted sources and signal

17:13 Models and document extraction

18:20 AI Engineer New York

20:32 Customer meetings

23:54 Hiring and customers

25:55 swyx’s path into AI

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit https://www.latent.space

Alex Lieberman and CJ Hess join swyx at AI Engineer New York to discuss Tenex’s approach to AI-native engineering, paying for outcomes, MetaHarness, enterprise AI adoption and rebuilding SaaS.

Show notes:

Tenex: https://www.tenex.co/

Building an AI-native SDLC: https://www.tenex.co/blog/building-an-ai-native-sdlc

Earlier Tenex episode: https://www.latent.space/p/10x-ai-engineers-with-1m-salaries

Kill My SaaS: https://www.youtube.com/watch?v=zlD-CkLVKtU

Full video: https://www.youtube.com/watch?v=Vki6cklFqxo

0:00 Cold open

0:26 Introductions

0:51 What Tenex does

1:32 Why CJ joined Tenex

3:17 Paying engineers for outcomes

7:20 Inside MetaHarness

10:25 Models and long-running tasks

15:11 Token spending and ROI

19:29 The enterprise AI adoption gap

23:22 Why agent deployment is hard

24:42 Vertical AI and enterprise software

30:32 Kill My SaaS

35:23 AI and jobs

38:18 New York’s AI ecosystem

42:28 Hiring at Tenex

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit https://www.latent.space/subscribe

This was recorded on the sidelines of the new Compute Trading track at AIE NYC. Catch us next at AIE CODE, our last flagship for the top agentic engineers in the world.

In the early days of cloud, former guest Martin Casado made something so disruptive that Cisco was threatened enough to pay $600m for it, and VMware outbid for $1.3B:

That concept was software-defined networking, which virtualized the routing network by separating the control plane from the data plane. That revolution is now coming for GPUs - where previously running your own supercompute cluster required $50-100m upfront to get commitments and do the entire stack from hardware to firmware to kernels, SF Compute is now enabling you to not only buy or sell GPU hours now and in the future, but also to programmatically control everything from your network to memory.

We last spoke with Evan about commoditizing compute. At AI Engineer New York, we went deeper: what does it take to physically deliver a GPU contract, and what has to be true before a market in compute can become liquid?

From selling a room to running the hotel

Evan describes SF Compute’s expansion with a hotel analogy. A customer wants a room, then asks for cleaning, room service, and security. Eventually you discover that delivering the room means operating the hotel. SF Compute began by trying to sublease a cluster; the company has since built much of the operating capacity of a cloud to deliver the compute traded on its market.

That work is physical settlement: the buyer gets usable compute, rather than simply a cash payment against an index. It brings the physical constraints of a cluster into the market itself. Can a workload be interrupted? How quickly can it recover from storage? Which GPU generation, networking and software configuration does it need? Those details affect both the contract and the price.

Making compute tradeable

Rather than waiting for every GPU cluster to become interchangeable, SF Compute matches orders with different constraints. Evan argues that this makes trading possible while the technology changes, at the cost of spreading liquidity across more variants. We also discuss spot and forward pricing, batch versus live inference, and why an index can become disconnected from the cost of actually buying compute.

Independence is central to Evan’s case for a market operator. He argues that companies with large hardware or inference businesses have incentives that make it harder to run a neutral exchange. We also get into the unglamorous engineering: visiting sites, designing clusters, running the backplane, and making infrastructure programmable. Near the end, he describes the ability to subdivide InfiniBand fabric on the fly as one example.

Guest: Evan Conrad, CEO of SF Compute.

Show notes

* Evan’s AI Engineer NYC talk: How Compute Markets Work

* Our previous SF Compute conversation

* Why oil futures went negative in 2020 — EIA

* How SF Compute customers resell unused GPU capacity

* SF Compute’s $245M compute sale

* Abel: SF Compute’s settlement ledger

* CFTC: Barclays and LIBOR manipulation

* Elastic InfiniBand partitions

Timestamps

* 0:00 — Why can’t NVIDIA build this?

* 0:25 — Welcome back to Latent Space

* 2:23 — From selling a room to running the hotel

* 3:47 — Physical settlement: actually delivering compute

* 6:00 — Liquidity and the GPU order book

* 8:42 — Trading nonstandard compute

* 10:00 — Interrupting and moving workloads

* 12:20 — Why hardware cohorts matter

* 14:32 — How far out do compute contracts go?

* 16:19 — Batch versus live inference leases

* 17:58 — The shape of compute prices

* 20:33 — GPU generations, specifications and SKUs

* 23:30 — Physical versus cash settlement

* 27:00 — The emerging compute market ecosystem

* 29:03 — What it takes to deliver a GPU cluster

* 32:24 — Sovereign compute

* 33:26 — Why NVIDIA can’t build this market

* 34:48 — A programmable data center

Transcript

Automated transcript, lightly formatted; speaker labels have not been verified.

00:00 NVIDIA is obviously like the 800-pound gorilla in the room. Why didn’t they make SF Compute? So our impression is that NVIDIA can’t make it, but benefits tremendously from it existing. The reason why they can’t make it is because they would then not be independent enough for people to trade on it.

00:25 Anyway, Evan, welcome back to Latent Space. Thank you so Welcome back to AIE. Actually, first time at AIE. I think so, yeah. This is a very special one because I feel like you have been working on compute trading, compute markets for a while. Now it’s like feeling of the moment. Yeah. Jump basically announced that they are getting into compute markets, which is why I’m excited to sort of cluster people. You met Oliver from SemiAnalysis, who is basically initiating coverage on compute markets is my read on this. So people are going to watch your talk.

01:00 I think it’s a very good explanation of both your origin story and why you need to exist and all the hard problems you work on. Yeah. Done in a very simple, engaging manner. I’m very jealous of your writing skills because you showed the flyer that you showed at New Reps. Yeah. I was like, it’s pretty straightforward and simple and it’s good. Yeah. So anyway, congrats. Thank you. I appreciate that. And also congrats on AIE. It seems to, y’all are rolling. We’re rolling. I just came back from the site tour for next year’s AIE. then for You have like a hint? At Javits Center. It’s the same place they do Comic Con. Okay. Wow.

01:35 Wow. Wow. Damn. But I think the finance focus is really hitting for folks. Yeah. So we’re going to try to split out. Like this is, here we just lump everything in finance. Like your finance, their finance, like insurance is finance, but like they’re all super different. Yeah. So we’ll probably break out things more minutely in future because people want to go deep on hedge fund stuff or banking or personal wealth. Yeah. Especially in New York. Yeah. Anthropic was like, yeah, we’re all about RIAs right now. Yeah. RIAs like registered investment advisors. Yes. Like so different from you.

02:10 No, no, no. I’m aware of this. Yeah. Yeah. Okay. So for listeners who don’t know, we did a podcast two years ago where you talked about your, you know, where SF Compute was. Um, today you’re, you’re much more in a sort of physical settlement. You’re talking, you’re talking about your, your back fabric or what do call it? Uh, the backplane. Backplane. Yeah. Which obviously that takes a lot of hardcore engineering stuff. Yeah. Um, can you give a status update on like, where is SF Compute versus listeners who last heard you on a podcast two years ago? So, uh, when we started, we were this company called Junark and then we were just trying to sublease our cluster.

02:45 Um, and we started there in the technical stack. Um, and a way to think about it is it’s like you’re in a hotel and you’re trying to sell that hotel room. And then somebody asks you, Hey, like, are you going to clean my room? Um, and then you say, Oh, I guess I probably also need the hotel room as well. Or like I need the, um, the cleaning closet. And so you go get access to the cleaning closet and then they want room service. And so you get access to the kitchen and then, you know, they want, um, security. And so you get access to the security closet and you get this and you do that and you do this. And at a certain point you’re just running the whole hotel. Um, and the moment you’re running the whole hotel, you say, Oh, wait a second.

03:21 Um, instead my supplier on this market, um, is somebody that looks closer to a, fund, um, or a cloud that has access in our case, a cloud that has access to the data center. And we’re going to level up. And so what we’ve spent the last several years doing is building the entire operating capacity of a cloud, um, in order to offer, in order to actually deliver, uh, the compute at a reliable level. Um, we call that physical settlement. We’re taking a term from the futures market, from futures markets, from other markets, um, where in, you know,

03:57 oil or soybeans or something, physical settlement means you get delivered the oil and the soybeans. Yes. Um, but, um, in some markets, um, like in oil and also in power, um, there’s like a delivery network in oil. You have to have pipelines. If you would like the oil to be delivered to you, there has to be a way for it to get to you. Yes. Um, and finance, and for people interested in financial history during COVID oil went negative because, uh, so like I will pay you to take oil from me because it is sitting there in the port. Uh, and like, I can’t pay for this.

04:32 Yes. And so I will pay you to take my oil. Yes. That is like the weirdness of physical settlement. Yes. There’s a bunch of weirdnesses like this in other markets as well. So in the power markets and the grid, um, there’s parts of the orders that you fill out, the buy orders and sellers that, um, are there because the boiler takes a while to cool off. Um, and you can, you know, stop the boiler and then it will, you can start it back up again within some period of time. And so that has to get factored into the order book. Um, this genre thing is the same thing that happens in compute markets. So we co-designed physical settlement and the order book at the

05:10 same time. And that meant that we get to actually handle the real constraints of compute, which might include something like, oh, I’m okay with being interrupted, but only if there’s like storage on the thing that I can recover from really quickly. Um, very similar types of problems. Yeah. So, uh, people can refer to your talk and you’ve done a couple other podcasts already. Um, so I want to be like the sort of 201 level instead of the 101 level. Sure. Uh, and also because I, I am, I have been a derivative trader, so I can give people a bit more detail that you normally don’t get into. Um, so for example, well, how, how mature or deep or liquid is the compute

05:45 market today? Uh, not very mature or deep or liquid. Okay. so what we’ve been doing is first we had to, we had to go get a bunch of supply online in the first place, which just takes a while. And then we had to build the entire platform in order to do this. Um, and we just now got that working. And so now we move, um, capacity over onto the order book, um, which makes it liquid. Um, and so 5% of our total capacity is on the order book and we’re consistently moving more and more. And then we’re bringing on more capacity. That’s all on the order book. Um, the nice part is that, um, maybe I should have said the, the volume on the actual order book is small, but the,

06:21 um, the liquidity on the order book is high. So for that small amount of capacity that you have, um, you have, um, like very concentrated liquidity. So, um, a sort of standard metric, I would say for someone who is using this is a customer who buys capacity from us and they spend a hundred dollars to buy that capacity. We’ll make back $25 on average from selling it back to somebody else. Um, and that’s the number that we, or this is one of the metrics that we try to track because, we want the liquidity to be useful to the commercial participant and you can measure liquidity in a bunch of different ways. You can say like, I just want tons of makers or something.

06:56 Um, and you can measure it in like high-minded ways. We try to measure it by like, is this solving the problem, um, for the customer? Yeah. Got it. And so, uh, last time we talked as well, there was a lot of contract, uh, conversation about standardizing the contracts. Yes. talking with like, uh, the CFTC or whatever, what’s the status of those things? Um, so I think standardizing the contracts in general is still pretty far away. Um, and the, what we were trying to do at the time was, um, or basically what we learned over the last several years was we just needed to

07:31 operate the whole thing. Otherwise you cannot standardize the thing. And so to get closer to something that can be standardized, you first need to make all the operations of the thing the same. Right. To give an example of what can be different is a single operator can offer two different types of virtual machines. One type of virtual machine, um, might have a 40% performance, um, degradation versus another types of virtual machine. And 40% is a lot. When you say performance degradation, are you talking like MFU type things or like amount of burn in that you’ve experienced or what? It will affect. So in this particular case, I’m talking about GPU kernel launches and how long it takes for that to happen.

08:07 Um, which is a process that happens quite frequently when you’re doing sort of either training or inference. Um, the, um, uh, that will basically, the end result is that you get a worse MFU, but, um, like is the intermediate level. the intermediate level basically. Yeah. Yeah. So, so what I’m trying to, so for anyone who’s traded commodities before, there’s different grades of commodity. Yes. Right. There’s like, it’s not oil. It’s light sweet crude versus like whatever the opposite of light sweet crude is, or it’s like, you know, uh, corn level a or corn B or whatever. Uh, and so like people are trying to, well, you have to establish the equivalents of those things. Correct. Yes. I’m wondering if they have been established.

08:42 So what we did instead, um, was we created an order book that had multiple different constraints and you can say, I want to have, um, you know, something that’s between this and that. Um, it runs from this time to this time. Um, it’s like, uh, X to Y, it’s a bunch of different properties that you can attach to your order. And then we match that against sell orders that have those properties, um, in a big matching engine. Um, and the reason why we did it this way is it allows us to generate liquidity without having to standardize before, know, what the technology can do. Um, and because the technology was shifting so quickly, um, we wanted to basically be able to do that. The trade off of this is that you have less concentrated liquidity.

09:18 And so the prices are going to have, there’s going to be more spreads between the, the buy and the sell. Yeah. Um, it sucks. Yeah. But on the other hand, it means there’s lots of arbitrage to make, um, if you can do that correctly. So if you have a product that can use lots of different types of capacity, um, then you might really like SF compute because you will be able to use lots of different types of capacity and you’ll make more money. Um, you can take the spread. Okay. What dimensions matter in this grid of things? Um, so a bunch of different things. Um, one is dimensions categories that I will say is like, is there a

09:53 workload running on it right now or is it capacity in the future? Sure. is it capacity right now? Um, capacity right now in order for you to sell that, um, you have to interrupt your current workload. So you have to think, do I care if my current workload is interrupted or not? Capacity that’s sold that’s starting a month from now, um, you can buy and sell that all the time. And so you get, um, liquidity on that really easily. Um, there is some concern like, oh no, well I have capacity next month, but it’s very different than if you’re interrupting a workload right now. Um, the way that the interruption happens matters. So an example is if you’re on a cluster and the cluster doesn’t have as amazing storage as you want it to have. Um, so an example is like, do you have a Weco or Vast system under the hood?

10:31 Um, and do you offer that? And like, what’s the size of that? Um, this changes how you respond to, um, being preempted or removed off the cluster. meanwhile, the location of that storage matters. So, um, if you’re able to move that data from one cluster to another, um, quickly, then maybe you might be able to spread across multiple different clusters. Um, and you can be interrupted over here and then pop back up over there. Um, to make that work, you also need, like a, a fast transit path between one cluster and another. Typically this is implemented with a WAN, but most of these clusters don’t

11:06 necessarily have, um, like a wide area network. Yeah. A wide area network. So, basically cables that connect the clusters together. A private internet is another way to think about it. uh, a lot of clusters that are being spun up today are like, they’re not attached to, um, traditional CPU clusters were, next to the customer for latency reasons. In GPUs, they’re next to the power, um, wherever the, like, you know, you put the data center next to the power rather than next to the customer. Um, this means that a lot of stuff is getting put in places that previously they weren’t expecting data centers there.

11:41 And so you may not have, um, as, um, excellent of a transit path, um, from one to the other. So this, this starts to matter. Um, other examples are, um, weird s**t where we theoretically support, but don’t actually, um, uh, like this is a thing that we’re thinking about, but haven’t done basically. Um, the early GB 200s broke. The early H 100s broke. Um, and so if you were buying H 100s or you’re buying GB 200s, you might really want the ones that were deployed after a certain time because you’re not getting the first shitty ones off the rack.

12:16 And so if you’re a super sophisticated customer, you may be willing to pay more, um, because for that. Oh, so it’s, it’s about like what cohort they were, they were produced and they get better over time. So like, uh, the first ones off the rack maybe had some faults or some problem. Right. And so if you were the ones who got the first ones, um, or, you first ones from, um, the, the first shipment of them, um, then yeah, they would, they would be broken or, um, in some way. I assume the sort of process is like, look, there’s defects everywhere. Like it’s, it just happens. Right. Totally. And so like you get it, you get a new shipment offer from TSMC or whatever, and you just kind of run your tests and like you, you figure out

12:53 which ones, which ones are bad and you throw them out. And isn’t that how it works? And I, I didn’t know about this element of cohorting where like, let’s say Q4, 2026, uh, B300s are way worse than 2027 B300s. Yeah. So, um, I’ll give an example. Um, you’re describing burn-in. Yes. so when you get a cluster, um, a good cloud, what it does. Is it tries to kill all the GPUs, um, before they ever get in the hands of the customer. Um, and then, um, the GPUs that are broken or the nodes that are broken, um, you get RMA’d. You basically go and request it. Yeah. You return it and you get a new one and there’s a lead time to that, but

13:30 whatever, you’ll get, you’ll get some back. Um, and then the customer never has to experience this. Um, the, um, thing I’m referring to is that, um, like in early H100s, there were like firmware bugs that may not even be related to the specific, like they may not be NVIDIA. They might be like Super Micro has a firmware bug in the, um, in the box that you bought. Um, and it applies to all of them and it doesn’t matter. Like, um, the solution to that was you need to get them to update their firmware. Um, or they need to ship a new thing, um, in every new box.

14:07 Yeah. had this for like the, the B300s that, that we got. Um, and the IQ had to use Devin to go fix it. Yes. Great. So like different types of problems like this can happen. Interesting. Okay. So I’m like, okay, look, that seems small compared to the overall. Like, I just want to, like, I want people to understand that the shape of this market, um, you mentioned like preemption, you know, and, you we talked about spot instances and how, how that, uh, the analogy is. Yeah. And then there’s also, you know, booking a month out. How far out do this, do these markets go? And is, is real time or sort of preemption more interesting to you versus,

14:42 far being further, further out? Um, to answer your first question, three to five years. Um, wow. We’re starting to see. month by month or quarter by quarter or how’s this? Are you asking how the cashflow works? Like the, the contracts, like the, the futures contracts, right? Like again, like I’m a former contracts, uh, derivatives trader. In the, in the market itself, um, like is in the broader market. Yeah. clusters are financed three to five years. Okay. Um, so the, the, the payments are totally over that period of time. Yeah. there is also like blocks of capacity in our order book, um, which can extend any arbitrary time. Yeah. there’s some stuff that we have to do, um, for, um, the cashflow on that

15:18 right now. What we do is we just ask people to pay entirely upfront for a block. Um, we don’t ask them to do that for a lease for a longer term contract. This gets into something that I definitely need a whiteboard to explain. Um, but, um, you had another question I was about to answer. I forgot. do you want real time or do you want, like, do you want more, uh, what’s more interesting to you? Real time or long, longer out? So we care more about longer outs because we’re trying to build a forwards curve that has a lot of liquidity on it. Um, the, um, these are kind of the same thing and kind of not. Um, so what we do is we, since

15:54 a, you can buy an hour out, you can buy right now to the next hour or you can buy right now to the next three years and anywhere in between. That’s wild. I know. Um, and so the, um, you know, or if I, you’re technically, um, the, there’s a certain difference that happens that’s dependent for each customer. Basically like if I have a workload that is, um, uh, that’s batch inference basically where I can be interrupted at any point. I don’t really care. Um, then my ability to be

16:29 cut off, um, is probably like, I need an hour of time to know, um, that I’m going to be cut off and then I’m good. Or I can even be notified in, you know, a minute or two. Um, but if I have, um, live inference customers who are expecting me to be online, um, then you can interrupt me, but in a month, maybe two months, maybe three months. Um, and then I can adapt to that capacity over that period of time. Um, but you cannot interrupt me in a minute. Um, and so those two things are priced differently. And the reason why they’re priced differently is because what is the price

17:05 you would pay in order to interrupt your batch job? It’s very different than the price you would pay to interrupt your, um, live job. If you have a live job, you’re, you are probably paying an SLA when you break for your customer. Um, does that make sense? Yeah, I see. So this is actually the market for breaking leases. These are the market for breaking leases. Oh my God. And then there’s also the market for future leases or like for leases that are, um. Oh, I thought this was like a subletting situation. This is like a lease break. So one way to think about it, um, is in power markets, there’s actually two markets. Um, there’s like the day ahead market and

17:40 there’s the live market. Compute kind of has the same thing, um, where it has capacity that is coming online, um, or capacity that’s about to, um, arrive to you. Um, and then capacity that is currently in use. And these have different price dynamics. Okay. Interesting. backwardation or contango? Backwardation or contango? Yeah. so do, uh, is compute further out price higher than spot or lower than spot? Um, is compute further out price? Does it, does it go like this? Does it go like that over, uh, you know, over like one year, one year out, two year out?

18:16 Do. Oh, this is an, uh, both at different times. I find it fascinating that you don’t know. Yeah. I mean, both. It depends on who you are. Um, okay. So, uh, I’ll make the case. Um, wait, so you’re asking is compute sold right now? same unit of compute today. Yes. Uh, maybe not today, maybe a month out versus one year out versus two year out. Is that, is that curve going up or going down? We see, like a hump. Oh my God. okay. So explain, um, like, I think this will make intuitive sense is basically like if I have milk and it is about to expire, I will lower the price on it. Yes. Um, and so that’s basically what we see.

18:52 It’s like, it’s the Nvidia release cycle basically, right? Uh, well, we’re more saying like if there’s a block of compute and it’s approaching now, um, it can’t be sold in the past. And so it has to drop its price in order to clear at some price. Yeah. And so initially you see this like drop where, um, stuff that’s sold right now, anything that’s on the market right now, um, is something that hasn’t been sold. And so therefore the price will drop. Um, and then it goes up, um, to where it wants to be. Um, and then maybe compute in the future is worth less, um, because people don’t want, like people would like the compute now. And so they’re sort of paying for the shorter lead time. Um, and they pay less for the compute that’s off into the future.

19:28 This is what I think a normal healthy curve looks like. Yeah. The place where I’m like, oh man, no s**t. Um, is that, um, right now if you took a B300 and you trying to run inference on it, um, you can make $40 an hour or so, um, selling that as inference. Any model basically. Um, and so that’s if you use it at a hundred percent and maybe if you only have like 50% utilization or something, so like $25 an hour. Yeah. So $25 an hour. Meanwhile, you can buy B300s at like five to $6 an hour. One of these numbers has to change. Yeah. It’s a huge arbitrage. It’s huge arbitrage. this in your talks a lot.

20:04 Yeah. So, um, a bunch of people are basically looking at this and thinking, um, if I think prices are going to go up, I might be willing to pay more than what I previously thought in the future. Um, because I might think, okay, I can buy a five right now, but if I can’t buy anymore, then maybe I will be willing to buy even more in the future just to lock it in if it’s going to come online later. So we see the other side as well. Um, kind of depends on who you are. Cool. So, is it okay. And, and I just, I just really want to make super clear on this. Of course. Do you care about the different generations of chips and, or do you just care about overall flops and you spread it among whatever chips you want?

20:40 Uh, we care about different generations of chips because. you care. I think like, I only, I guess do your customers care? Yes. Right. Oh, wait. When you say a different generation chips, do you mean like. So like B300s, B300s. exactly. Yeah. Or like even AMD or whatever. Right. Uh, yes. I think people totally care about the generation of chips. Um, uh, I think part of this is because of things like VRAM. Um, and part of this is because of like, um, larger NBLink domains. Yeah. And things like that. So there’s just no way to abstract all this to like some generalized unit of compute that you can sort of mush around and whatever you like. I think like when you were saying oil, even oil isn’t oil. Oil is actually just like a bunch of different types of oil.

21:16 Yeah. Like jet fuel or car fuel, whatever. I think it’s the same thing. Um, you’re going to get a bunch of different types. It is, but you have oil refineries. Yeah. can effectively have black boxes. You can put oil in and get like whatever out. Um, if compute, it is so heterogeneous, I guess is the word. It’s that you just have to expose the underlying complexity to your end customer, which is bad for you as a market maker. Maybe. Like you want to identify the one like choke point or like, I don’t know what the better word for choke point is like a shelling point where everyone agrees like, okay, like this is kind of the standard and we’ll, we’ll all agree to this standard and, uh, you know, it’ll

21:53 be somewhat inefficient. Yes. Uh, so we’ve been able to reduce the, um, the, the problems to a certain level, but yeah. Like your job is like dimensionality reduction. Yes. We reduce a lot of the dimensions, but nonetheless, there are still dimensions. Yeah. It’s still going to exist. Do you, do you know how many SKUs you have? So, um, like five times a hundred times like 5,000 equals, I don’t know. Like what’s, we actually made SKUs like a concept, um, that, and the goal is to make it so that somebody else can make their own SKU. So if I want to buy Blackwells, I can say I’m willing to

22:31 buy Blackwells at $6 an hour or less. And you can match me against GB300s or you can match me against B300s and I don’t care which Blackwell as long as it’s less than $6 an hour. but you just said you care about memory and other stuff. Totally. But not every customer does. So some customers like, um, just can work on Blackwells, whatever it is. Some customers are super hyper specific and they need this exact thing and they’re going to pay more for that because the market’s going to charge them more. Okay. Um, and that’s what we do. So, and so people use your APIs to programmatically define what they want and then you match that with whatever composition of sellers, uh, they might, they might, they might be able to fulfill.

23:08 Yes. Um, with a caveat that having a named query is a new thing we’re doing. Um, but the query system as a concept is how the order book works right now. Okay. Interesting. Um, all right. So, uh, you know, I, I think that’s like a lot of like, I guess, nitty gritty. Are there anything about like, I guess that you want to educate people on whether it’s financial people or engineers about the state of the compute trading market as, as you’ve developed it? Because effectively you guys are defining it. Um, mainly just the requirement for physical settlement. Yeah. Um, we think that if you, we, we spent three years learning why you need

23:44 physical settlement. Um, and, um, our entire conversation today is, um, that it is, these things are not as fungible as you think we’re going to try and make them fungible. That’s the industry’s job to do that. Um, but without that, you run a lot of risks. Um, if you try to build, um, too much financialization before you’ve, the industry has had time to build out the, um, the physical settlement. Um, I think what you have is you have, um, a market that’s very easy to manipulate and a market that has a lot of basis risk where, um, the price you may be hedging against actually has nothing to

24:22 do with the price that, um, the market, you can actually buy the capacity for. Yeah. So for, listeners, uh, we, we talked about this briefly last time, but, um, people are often shocked to find out that let’s say the market for credit default swaps is 12 times the, the, the size of the bonds that they insure, uh, something like that, or like, uh, the, the markets for mortgage, um, sort of reinsurance or, uh, all that is, is larger than the underlying physical settlement. So that’s why physical versus cash settled. Yeah. And, and like basically cash settlements are like exactly the kind of, uh, the side bet that you see in the big short, um, where, uh, they,

24:59 they may, uh, may or may not have anything to do with the underlying thing itself. And it may, may be used as a proxy for overall other stuff. Yes. That, and you don’t want to, that to go out of control. Yeah. So to be clear, we think, um, cash settlement as a future concept is a great idea. Yeah. Um, we just think it should be derived from physical settlement. Yeah. because as you said, it’s a side bet. Um, and it would be kind of like if you, I mean, be kind of like if you made a cash of the future before you made the physical settlement, um, which I think people are kind of doing already. Yeah. Totally. People are trying to make cash on it before you make the physical settlement. And people do that for all sorts of things, including

25:34 startup stock. Yes. Right. Like you can buy and sell a made up share of open AI, but it’s nothing to do with the actual share. Yes. Um, and lots of those go very poorly. Um, like it would be kind of bad if you made a, um, it would be bad if you started selling, um, the, the derivative on a stock that doesn’t exist yet. Um, yeah. Yeah. Well, so I mean, one thing that we do, so I used to be a currency trader, right. Yeah. And, interest rates as well. And what, one thing we do is you have a board that gets together and says like, this is the London interbank offered rate LIBOR. Yes. And we all agree like this is the, the, the monthly fixing and we all

26:12 trade to that fixing. Yes. Uh, is that even meaningful at some point in three years from now? So critically LIBOR was manipulated. Yes. our concern at the moment. there’s 12 banks? Yeah. Yes. Um, well, our concern is there’s a very similar problem, um, where you can manipulate an index. Um, our impression is why you can manipulate that index is because a lot of the indexes are based on a very small amount of data that has, um, uh, not a lot of correlation with the actual price of buying a new reserve contract. Um, that’s what we’re concerned about. Okay.

26:47 Interesting. Um, all right. I think, I think that’s the, the, the rough state of things. Uh, um, I think the other part of just, just today is, uh, you, you, you, you talked about an event with jump and you’re, you’re friends with Silicon data. Uh, what are you learning from all these other players? Like who are the major players that you would want to call out as like doing interesting things? Ooh. Um, so we’re big fans of Silicon data and compute exchange. Um, I think they were also very early to this. Um, and I think they are doing like a reasonable shot at the, a path that we think, um, that we’ll likely end up being partners with.

27:23 Um, can you distinguish like your, your different paths? So, um, the general just is they’re doing cash settlement. We’re doing physical settlement. Um, we think both can exist and should exist. Um, and that they’re sort of mutually beneficial to each other. Um, and so totally fine. Um, so I think they are one of the earliest in the space and I think they are, very clearly took the bet, um, well before anybody else did. And I think that’s like very honorable to do. Um, um, you mentioned, uh, jump. Semi-analysis, jump. Like these are the players emerging. Like as far, like you were my first introduction to this. Yeah. And then, and I see all these other guys coming in and I’m like, okay, like it’s something’s happening here.

27:58 I don’t know the landscape as clearly as you do. Yeah. So, um, the categories of folks that we’ve seen is there’s folks that are trying to set better standards. So semi-analysis I think is working to try and create better standards. Um, there’s folks that are trying to do, um, residual value insurance and so on. Um, there’s a number of different companies in this space. I don’t know any of them super well, but, um, I know, know this is a concept and I think that’s important. Um, there are folks that are trying to do, um, uh, deposit, um, lending where like they’ll front your credit risk for the cloud. There’s an ecosystem of companies that are forming around how do we resolve the financial bottlenecks of compute.

28:34 Um, and it is cool to see. I think effectively what you want in this ecosystem is buyers and sellers. You want an analyst. You want, uh, we said insurance or, uh, what’s the, uh, residual value. Yeah. Yeah. and you, we want this like healthy ecosystem out there. Um, but obviously you guys are owning the problem of physical settlement, which, which is really interesting. Um, okay. So, I mean, we can go into like the, the hard tech side because obviously you are also hiring, uh, for extremely cracked engineers. Yeah. describe your stack. Um, we do all the things a cloud does without being the

29:10 cloud. Um, as in we don’t own the hardware. the reason why we don’t own the hardware is because if we did, we would be able to manipulate the forwards market. So we have to be Switzerland. We have to be independent. Um, which creates a lot of sort of strange things about the company. Um, but in order to do physical settlement, you have to do site visits. have to visit the data center. Um, you have to, um, you have to design the bill of materials and ensure that, um, you have to basically do like supply chain consulting, um, where you have relationships with different OEMs that, um, ensure that things get delivered

29:46 on time. Um, then you have to, um, do cluster architecture. Um, uh, so you’re sort of designing the cluster itself. Um, then you’re, um, rolling out racks or you’re working with partners to roll out racks and supervising that. Um, then you’re doing, uh, backplane management. So you’re taking over the, um, uh, PDUs and the switches and the DPUs and, um, UFM and BMC and you’re operating, um, the whole, um, harness around the nodes that you’re selling. When you say you as an SF compute employees. Yeah. Wow. Um, so, uh, when you buy a node from a cloud, um, there’s the bare metal, but there’s all this s**t around the bare

30:23 metal. Yeah. Um, and in order to make that bare metal node work, you have to operate everything else. Then we. Because I just like traditionally the job of the data center is to extract some of this away from you. Yeah. We’re the ones that do that. Okay. Right. Um, but you’re not, you’re not in the data center business. You’re not the sort of equinixes of the world. We’re, we’re not in the data center business sort of, um, what we do work with is, so we offer supercomputing as a service. Um, and what this means is folks come to us and say, you know,

30:58 I’ve got X where X is, I’ve got, you know, X megawatts here. And maybe sometimes they mean I’ve got a data center. Um, that’s like, um, would like to turn into a GPU cluster. Like my options are, I could sell the GPU cluster or I could sell the data center to somebody or I could become the cloud. And then we go to them and say, great. Um, we will turn you into the cloud. We’re the supercomputing part. You’re the data center. Let’s partner together. And then we will sell and I’ll make money. Um, sometimes it’s people who have something that could turn into a data center, in which case we’re going to basically project manage that up to the layer.

31:33 Um, and then we will work with those far, we’ll work with those folks to turn their powered land, um, or whatever they have, um, into a data center by working with partners. Um, and then we take over at the supercomputing part. Okay. This requires us to have some sort of data center expertise in house, but we don’t self-perform builds. Um, so yeah. Yeah. So it’s, uh, somewhat modular because you have different scenarios you need to play under. Yeah. Uh, someone once described us as like, um, uh, like Palantir for compute, um, because it’s like these, each time it’s this new rodeo.

32:10 Um, uh, but, and we like, we, we deeply partner with folks. On both sides. Yeah. Uh, and primarily the U S primarily the U S, but we are operate around the world. Yeah. At some point does SF compute become a misnomer? Uh, so when, so we’re deploying clusters all around the world. Yeah. And when we do that, we’re working with people who themselves are trying to be clouds all around the world. Um, they look like sovereign compute players. Um, and our pitch to sovereign compute is amazing. Uh, let’s help you set up a cloud in your own country. Um, and we will bring the San Francisco offtake. We’ll bring the San Francisco expertise.

32:45 We’ll set up the clusters. Um, that’s our whole business. Ah, um, we’re going to make it. Who are your main, uh, your biggest announced partners that you can name? Um, nobody I can name. Really? Wow. Uh, we will have some announcements, um, in a week or two, um, that we expect to come on. Yeah. All things sort of. Sure. Sure. Good luck. Yeah. No, I mean, uh, I, I think this is really interesting for every country to like, who’s, who’s like investing in sovereign AI to figure this out. Um, and it’s interesting that like SF is still sort of like that brand name and, uh, that, that is, that is very marketable.

33:20 So I think one of the interesting things is just thinking through this market. NVIDIA is obviously like the 800-pound gorilla in the room. Why didn’t they make SF Compute? And again, the answer is going to be market manipulation. But let’s just really get into the incentives of the multiple players. Yeah. So our impression is that NVIDIA can’t make it, but benefits tremendously from it existing. The reason why they can’t make it is because they would then not be independent enough for people to trade on it. If you don’t trade on it, then you don’t have enough liquidity for lenders to finance against the forwards market.

33:55 This is also why Crusoe and CoreWeave and Lambda, why the hyperscalers can’t make it, why Base 10 and Fireworks and Modal can’t make it because they have too much demand on that side. This is a thing that benefits the entire industry because it makes the financing costs go down. But nobody can make it unless they give up the ability to do all the big inference money and all the big cloud deals. You have to kill your business in order to make this. That’s why nobody does it. like structurally, they’re not set up to do that. And you’re basically reducing friction and that’s all your revenue.

34:30 You’ve announced some milestones. Actually, I forgot to say that up front. Can you repeat what they are and just take a victory lap a little bit? We’ve closed several hundred million dollars of volume. We have a significant number of new clusters on the way. And then the things that I think I’m most excited about is we’ve shipped a whole bunch of new features that make SF Compute very viable for platform-as-a-service companies. An example is we allow customers to subdivide

35:06 their InfiniBand fabric without having to ask us. So many clouds, when you go to buy them, they will give you one InfiniBand fabric. And then if you want to redraw the lines of that InfiniBand fabric, you need to contact them and they will then do something for you. We now can do that on the fly for customers, which I think is really great because it interfaces with the market really well where you want to be able to redraw things in order to scale up and down your clusters. Would you describe this as like you’re virtualizing the data center and making it programmable? Yeah, we’re virtualizing the data center and making it programmable. So all the stuff that you would have gotten if you had owned the

35:43 hardware, we’re trying to virtualize that, which I think is exciting to us. It’s funny, like the last time this happened was like the cloud, which is really cool. And then finally, you know, I think what is your call to action for people? You know, is it engineers? Is it like, give me your supply? Like, what is it? If you would like to own a GPU cluster and you want the residual value from it, you want to have it for your portfolio, we would love to work with you in order to build that GPU cluster. We will take over and we will do the supercomputing portions of that and

36:18 ensure the things go right. We own the SLA. And then if it turns out that that was a bad idea that you need out of that cluster, we can immediately sell that cluster to somebody else so it doesn’t hit your balance sheet. Very simple pitch. Well, thank you so much for your time. You’re doing a great service to the industry and I just am happy to be your friend. I appreciate it. All right. Thanks. That’s it.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit https://www.latent.space/subscribe

A few years ago I stumbled upon Leash Bio’s Substack, and was hooked. Be it their Kaggle competition that showed that no, we really haven’t solved protein-ligand binding, or their crazy discovery that knowing a molecule’s author is perhaps all you need to know to predict how good it is. Their substack — or at least Owl Posting’s deep dive on them — should be required reading for any scientist trying to do chemistry with AI.

So I was super excited to finally be able to sit down with Ian Quigley from Leash Bio, and pick his brain. Leash specializes in DNA Encoded Libraries (aka “DELs”), a way of collecting ridiculously high-throughput interaction data between proteins and small molecules. Before I discovered Leash, I had heard nothing but horror stories from DEL users. And yet, Ian and friends somehow found a way to tame the beast, extracting clean and reproducible data that is just perfect for large scale modeling. Without teasing too much, the answer comes down to epistemic rigor, and an incredible conviction that you just have to Do The Thing.

It gets even better! Near the end of this episode, Ian teases something his team was just playing with when we recorded: an AI playground for agents to do verifiable scientific reasoning. They discovered the data they had spent the last few years collecting could provide tens of thousands of “lab-in-the-loop” simulations, just what an agent needs for scientific RL. Two months later, it’s matured into Minos, and today they’re releasing a full preprint, blog, and open dataset. We’ll be releasing a lightning podcast in a few days with an update. Stay tuned!

We had a blast with Ian, probably one of our most fun episodes to date. For both complete chemistry novices and expert AI x Bio practitioners, we think you’ll get something out of this.

Enjoy!

We covered:

* Why DELs traumatize everyone who touches them, and how Leash puts dozens of barcodes on each molecule to catch proteins that bind the DNA instead of the drug.

* Ian’s rule that biology only scales two ways: taking pictures and sequencing DNA.

* The Kaggle competition: about 4B measurements from a basement and about 2,000 teams, with nobody beating chance on an unseen chemical library. Also why Leash quietly feared co-folding would make them obsolete.

* Clever Hans in ChEMBL: models can guess a molecule’s chemist from its structure, and “who made it” predicts activity about as well as the structure does.

* Hermes, trained only on Leash’s own data, keeping pace with Boltz on public benchmarks. Plus the BRD4 molecule Boltz had already seen in the PDB.

* Breaking the pocket: engineering protein mutants at scale after catching Hermes memorizing “nicknames” for proteins instead of binding sites.

* Lumacaftor-like hits as an accidental detector for misfolded proteins.

* The agent sandbox: LLMs that reason about SAR in plain English, hunting for the best binders among 6.5M molecules on a budget of 100.

* The Recursion mafia, and why its alumni companies all rhyme.

* The Kosuri paradox: if you can collect the data, why do you need the model?

* Going from a DEL hit to good mouse PKPD in six months on an inflammation program, and the case for pills over antibodies on cytokine targets.

* What stops a better-funded lab from outscaling them, and Ian’s by-fiat wish for massively parallel purification.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit https://www.latent.space/subscribe

Watch along AIE NYC and join us at AIE CODEin SF next month!

This episode includes a live demo of the Taku robot that is best enjoyed on YouTube.

Last month, Dyna made waves by announcing Dyna-2, a world-action model pre-trained on one million hours of human video, claiming several new scaling laws on human data across four orders of magnitude. They have now followed up the model with a physical robot agent, Dyna-2.1, that has to be seen to be believed:

In his talk at AIE this year, Jason talked about how the real challenge for robotics in the real world is reliable, long-horizon autonomy, and how Dyna got its models to a 99.4% success rate folding restaurant napkins for 24 hours straight, including recovering after it pulled over the whole stack.

From training a world-action model on over one million hours of human video to deploying robots that fold thousands of napkins daily at Din Tai Fung, Dyna Roboticshttps://www.dyna.co/is pushing physical AI beyond impressive demos and into real-world production. In this episode, Dyna Robotics co-founder and Chief Scientist Jason Ma joins swyx on-site to unpack how robots can learn new tasks from as little as 45 minutes of demonstrations, why predicting the future of the physical world may be the key to general-purpose robotics, and what it actually takes to build robots that work reliably for real customers.

We go deep on Dyna-2, Dyna’s million-hour world-action model that combines future video prediction with robot actions, the scaling laws behind human-to-robot transfer, diffusion transformers and flow matching, and why synthetic environments and self-generated training data could unlock the next wave of robotics progress. Jason also explains why Dyna builds its own hardware instead of relying on off-the-shelf humanoids, how agentic harnesses enable robots to execute hour-long workflows, why dexterous hands and missing sensory data remain major bottlenecks, and what separates real commercial deployments from the graveyard of robotics startups. We also cover world-action models versus vision-language-action models, GPT-6 and coding agents in robotics, and Jason’s journey from AlphaGo and Penn to NVIDIA, Google DeepMind, and founding Dyna.

We discuss:

* Why Dyna-2 was trained on over one million hours of human video

* World-action models: predicting future video to guide physical robot actions

* Diffusion transformers, flow matching, and real-time robotic control

* How robots learn new tasks with as little as 45 minutes of demonstrations

* Why robotics has no equivalent of the internet and data remains its biggest scaling challenge

* Human egocentric video, robot demonstrations, and cross-embodiment transfer

* Synthetic environments, simulation, and models that generate their own training data

* Why robots need on-device inference, model distillation, and custom hardware

* Why Dyna chose wheeled semi-humanoids instead of traditional bipedal robots

* The hardware bottleneck: why dexterous robot hands still overheat

* Best Western, Din Tai Fung, and the difference between robotics pilots and scaled deployments

* The Bitter Lesson of Robotics: scaling data instead of engineering specialized solutions

* Agentic harnesses and robots executing multi-step, hour-long physical workflows

* GPT-6, coding agents, and AI-generated simulation environments for robotics research

* World-action models vs. VLAs, and why vision alone may not be enough

* Jason’s journey through AlphaGo, Penn, NVIDIA, Google DeepMind, and the Eureka project

* Dyna’s research culture, deployment-first philosophy, and vision for millions of robots

Jason Ma

* LinkedIn: https://www.linkedin.com/in/jason-ma-742224a2/

* X: https://x.com/JasonMa2020

Dyna Robotics

* Website: https://www.dyna.co/

* X: https://x.com/DynaRobotics

Timestamps

00:00:00 Introduction and Dyna-2’s World-Action Model

00:03:08 Diffusion Transformers, Flow Matching, and Generalist Robots

00:08:19 The Robotics Data Bottleneck

00:13:50 Synthetic Data and Self-Improving World Models

00:16:51 Real-Time Inference and Model Distillation

00:19:45 Full-Stack Robotics: Hardware and Supply Chains

00:25:02 Why Dyna Built a Wheeled Semi-Humanoid

00:29:35 Teaching Robots New Tasks and Evaluating Performance

00:31:59 Autonomous Workflows and Hardware Limitations

00:36:06 Real-World Deployments at Best Western and Din Tai Fung

00:41:07 The Bitter Lesson of Robotics

00:46:13 Agentic Harnesses for Physical AI

00:49:07 GPT-6, Coding Agents, and AI-Native Robotics

00:52:02 World-Action Models vs. VLAs

00:59:17 From AlphaGo to Founding Dyna Robotics

01:04:20 Dyna’s Research Culture and the Future of Robotics

Transcript

Introduction: Jason Ma and Dyna Robotics

Swyx [00:00:00]: All right, we’re here on site with Jason Ma, the co-founder, chief scientist of Dyna Robotics. Thanks for having us in your office.

Jason Ma [00:00:06]: Of course. Thank you for coming.

Vibhu [00:00:07]: It’s been a blast. We got to see—you’ll see plenty of B-roll—drinks being pulled, a lot of evals, robots with knives for the first time. So under the hood of all this you have a world action model. What is that?

Dyna-2: A World Action Model That Predicts the Future

Jason Ma [00:00:22]: So last month, we released our world action model called Dyna-2. So it’s the first world action model trained on more than one million hours of data. So going back to your question first, world action model is a model that’s able to predict the future video, the future state of the world, so hence the name world model. But in addition, you can also use that future generation to also produce actions robots can execute in the physical world to control the robot. So why do we do that? It’s a very powerful architecture, because if you think about how humans act in the physical world, in our brain, we are implicitly thinking about how the world will evolve based on our actions. So now we have a model architecture that’s able to do that to guide robot actions. So what we find is that by doing so, the model is much more dexterous, much more robust, and much more capable of generalizing to environments, to scenarios it’s never seen before.

World Action Models: Inputs, Outputs, and Visual Reasoning

Vibhu [00:01:17]: I wanna prompt this question a bit. I think a lot of people hear world model, and they think you’re outputting video or world or 3D state. You’re also doing action. I think most people don’t understand what models are under the hood of robotics. So what are the inputs and outputs? What’s the training data? What is the architecture? Give us all the details.

Jason Ma [00:01:39]: So, for inputs, the most important input is video, right? You can think about videos that the robot is capturing with its cameras, right? So video or images of the past many seconds, that’s the input. But for robots, what’s really interesting is that robot also has a lot of actuators and joint encoders on itself. So the robot also knows, the angles of its joints. So those can also be inputs to this world model. That’s something that’s very different from just a pure video model that’s only looking at videos, right? So that’s the input. And additionally, for robots, you want to be able to talk to it, to instruct it to do different tasks, right? So you can also feed it language instruction. So those are what I consider the basic set of inputs that go into any robotics models, including world action models. And once we have that the output, at the basic level, it needs to output actions the robots can execute on the hardware. But in the case of world action model, what’s really interesting is that in addition to the actions, it can also output future video, right? So basically, you can almost output the future video or some representation of it first, and then use that to guide robot actions. So the way I think about it is this is kind of analogous to chain of thought, you know. And basically, the models are reasoning in the visual, the physics space, and then outputting actions. So that’s the key difference between world action models outputs versus other kind of models, robot models outputs.

Diffusion Transformers, Flow Matching, and Smooth Actions

Vibhu [00:03:08]: You put out a good blog post around your model. I think a lot of people understand language models or transformers, next-token predictors. In this case, you shared quite a bit. There’s diffusion, flow matching, you’re predicting next world state plus action. One thing you say is a million hours. How do you determine the architecture? Like, what are the choices there? What is the architecture? I feel like in language models, there’s a lot of take something off the shelf- Take trillions of tokens, do prediction, SFT, RL, and you have a model. What about this? How do you go from nothing to your model? Is there vanilla architecture off the shelf? Do you determine all this and what’s, what is the current state?

Jason Ma [00:03:49]: Yeah, I think, yeah. So the mental framework to think about how do we build models for robotics is that there are essentially a couple things to look at. One is data, and the second is how do we make a model learn as much as possible from that data? So let’s first talk about the model actually, right? So the blog post we released is our Dyna-2 world action model. So under the hood is a diffusion transformer. The model uses an encoder-only transformer architecture.

Vibhu [00:04:20]: Right, yeah.

Jason Ma [00:04:20]: But what’s really interesting is that in visual domains, right, if you think about generating images, generating videos, and in this case, robots looking at images to generate actions, people have found the diffusion kind of architecture to work really well. So what is a diffusion model? A diffusion model is a model that takes some random noise and through iterative computation can go from noise to something that’s really high-dimensional, that’s something really highly defined and have very fine-grained feature. If you think about generating images, right, you want sharp edges, sharp corners, color contrast. And for robotics, even though you’re not generating image, you’re generating actions that needs to be very precise, very smooth, right? So it has kind of the same kind of consideration and desiderata as generating image. So that’s why from first principles, the whole field, not just us, have been using this kind of diffusion transformer architecture to train robotics foundation models. You need the transformer aspect because transformer is the thing that can just ingest the most amount of data, can learn. Then diffusion helps you to refine the outcome to get to the smooth, precise, fine-grained motion you need to actually execute on the robot. So that’s the architecture that goes behind a lot of our recent breakthroughs.

Swyx [00:05:37]: And we’ve covered flow matching before on the podcast. It is mostly an efficiency speed-up. Like, let’s say you normally would take forty to a hundred diffusion steps. You can reduce it to four.

Jason Ma [00:05:48]: Yeah, that’s right. So with the diffusion model, one thing that’s very hard in practice is that you need to tune the scheduler, like by how much you’re tuning or converting the initial noise input into the final output. So flow matching is kind of like a special case of diffusion that makes the transformation much simpler. In practice, you don’t have to tune as much, and it works just as well. That is why I think about two to three years ago, most generative modeling for image and video has shifted from a diffusion model to a flow matching model. But the concept is the same, and flow matching is just a special, simpler to tune, variant of diffusion. Yeah.

Swyx [00:06:26]: It actually makes me think that. So flow matching obviously has trade-offs in quality. But actually, it’s very suited for action diffusion and robotics because you get to reevaluate in real time. Whereas in images, you are— it’s basically one shot.

Jason Ma [00:06:40]: That’s right.

Pretraining a Generalist Robot and Adapting to New Tasks

Vibhu [00:06:41]: Which is pretty fun. We got to mess with the laundry folding robot … throw a random towel in. The models adapt, right? I think one key thing about your models is that they’re tr— they’re made to be generalists, so one foundation model and you adapt them a lot. How does that work? Basically, in 45 minutes of mapped data, you can generalize towards. We saw an example of pulling drinks out of a fridge. What’s the what’s the mindset there of one general model, and then how do you adapt to different tasks?

Jason Ma [00:07:14]: So the way we think about it is that for robotics, we need to build a very general, strong base model, right? And then— Because if you think about how a human interacts and behaves in the physical world, right, we have very strong base knowledge. We can just readily do a lot of tasks. But if you want to become an expert at any given task, there is some amount of training that needs to happen. And I think to train a general-purpose robot like the one we’re building here, the same principle has to apply. So our approach is to first train a general model that’s training on as much data as we can find. So that is why in our Dyna-2 report, our model is trained over a million hours of human data as a starting point. And then given that strong foundation, what we found is we can quickly make the robot or make the model do many different tasks with a very minimal amount of fine-tuning to just tell the model, “Here’s the task I want it to do, and here’s very few amount of demonstrations,” and the model can readily do bunch of the tasks that you have seen in office. Yeah.

A Million Hours of Data: Egocentric Video and Robot Demonstrations

Swyx [00:08:19]: We actually have quite a few people who deal in data, who sell data. What is good data? What is bad data? What do you want to buy? I think that these are all open questions. Like, obviously, I think the top, important thing is egocentric data of humans doing tasks. What else would you modify on that basic statement?

Jason Ma [00:08:42]: So I think data is the most important thing in robotics today, because unlike language model, there is no internet, right? So every company that’s building model, building hardware has to think about data. So there’s, roughly couple of different ways of operating here. One is that you can collect a lot of data yourself, which is what we are doing as well.

Swyx [00:09:02]: You have employees here doing that.

Jason Ma [00:09:04]: There are robots everywhere in the office, right? But if you think about how much data you need for robotic foundation model, I think the amount of data any individual company can collect by themselves is just nowhere near sufficient, right? It’s kind of like, you’re trying to build an internet by just having Reddit or just having Facebook. They’re great websites, a lot of users, but it’s not enough, right? So when we think about data, there has to be some sort of data acquisition happening, right? You buy data, you look at data that’s already out there. And then there’s many different tiers, right? You mentioned data quality. I think data quality in robotics is a non-stationary problem in the sense that what is high-quality data for your model changes over time. Like, when your model doesn’t know anything, like when it’s zero, any data is useful, right? You can use a lot of video data. In the case of egocentric data, I think it’s actually a very good way of pre-training robots because human data is so diverse. Like, it covers every single physical activity you can possibly imagine. And that gives robot, the model such a strong prior on how things, how the world can evolve around a embodied actor, human. But the issue with human data is that until your robot looks exactly like a human, there’s gonna be some gap between human actions and the robot actions, right? So just by training on human data, today, it’s not very likely you can just execute robot actions directly. So you still need some amount of adaptation using the robot’s own data, right? But the robot’s own data collected on a robot is gonna be much more expensive and much harder to collect. So the kind of recipe we’re thinking about is like, how do we take the largest source of data we have today, like egocentric data, to give the model a strong prior over everything physical manipulation can cover, and then combine that with a small amount of robotics data to make the robot master whatever task we’re interested in. So data quality and getting data in robotics is really, I think it’s a fascinating, multifaceted problem. But the reality is that there’s not a single winning strategy like, “Oh, here’s one kind of data. I’m only gonna collect on that,”

Jason Ma [00:11:25]: Which will give you a strong foundation model. You really need to think about, “This human data covers this part of robot cognition I need. Robot data covers low-level reflex dexterity the robot also needs,” and put them into one model. And I think that’s what gives a really adaptable, really strong model at the end. Yeah.

Data Partnerships and Specialized Training Examples

Swyx [00:11:44]: I’ll prompt you, just for again, for the people listening who want to work with you. What, d— what data vendors do you want. This is your call for data. What-

Jason Ma [00:11:55]: Yeah, no, that’s a great question. And yeah, so in our Dyna-2 blog, at the end, we actually have an email address called datapartnerships@dyna.co. So if you’re a data vendor that has a lot of high-quality, let it be egocentric data, robot data, video data, or anything you think could help robots, send it to that email. Our team will take a look, and we’ll get back to you.

Swyx [00:12:16]: But at your stage, are you. You’re probably not buying any more egocentric data off the shelf, right? That you probably want to be like, “Here’s what I need. You go-”

Vibhu [00:12:26]: Is there a niche that’s lacking?

Swyx [00:12:28]: You need a Scale AI, you need a Mercor for doing this. I see. The custom data-

Vibhu [00:12:32]: The folks that don’t work in robotics are probably thinking like, “Okay, some niche factory.” Like … you showed putting tubes in. There’s only so much data you’ll find about that but like would a lot of that help you?

Jason Ma [00:12:43]: Yeah, no. If you really want to nail a task, having data from that task will certainly help you a lot, right? So for us, it’s like we still think, just a million hours of data is not enough. It’s just the beginning. Like, we showed the first model on a million-hour scale, but we need more than that right? So we are still looking at a lot of just raw, diverse data, that we can get. But we’re also thinking about how do we get more custom data either collected in-house or by our partner for really, specific niche tasks? Because I think you still need that to get to excellence that any model needs to actually be useful in the physical world, right? It’s kind of like humans, you know. If, Sean or me, we go work at Rosewood Hotel tomorrow, we are not gonna be able to make the bed the way they want, right? We need a lot of training. So that kind of data, like special data, that’s kind of like what Mercor is doing today for language model, I think will be very valuable for robotics as well.

Vibhu [00:13:42]: So hotels, strap a GoPro on your employees’ heads and.

Jason Ma [00:13:48]: And then there’s probably more of that.

Vibhu [00:13:49]: There’s the

Synthetic Data, Simulation, and the Eureka Project

Swyx [00:13:50]: Put themselves in their way. Yeah. So one thing that we discussed is the closing this interview, and what’s very interesting is now you can basically fabricate environments on the fly. Have you tried that with the frontier models? What’s your experience? Are they good enough?

Vibhu [00:14:06]: Are you using any world model robotic. Like, world model companies to generate these and train on them?

Jason Ma [00:14:13]: So, I think generating synthetic environments, synthetic data, simulation data for robotics, it is going to be one of the next frontiers, right? ‘Cause today people have looked at egocentric data, robot data. Those are scaling, but I think synthetic data will be a new dimension of scaling. So going back to your question yeah, we have been exploring using frontier models to generate synthetic, environments for us on our data, on our environments that we care about. And surprisingly, I think now it’s at a point where this technology is actually very much worth going really deep into. So for some context, so, in my PhD, I primarily focused on simulation-based training for robotics. Back then, simulation wasn’t ready, but we had this paper called Eureka, which was the first paper taking a large language model. At the time, it’s GPT-4, which is how to self-generate environment, generate reward functions or reinforcement learning, and just do things in simulation. And at the end, I think, we had this really nice demo of the robot spinning pencils, right? That was really cool at the time. But that kind of thing was still very limited at the time, because back then the model didn’t really have good vision capability, didn’t have a innate 3D understanding, right? So it was basically just generating code and using very limited feedback from text to try to come up with new simulation environments. But today the vision capability of these frontier models are crazy. They can do real to sim, do CADing, there’s 3D reconstruction and all of that. So I think now it’s at a tipping point where

Jason Ma [00:15:52]: This technology could really accelerate robotics. And for us, right, we think about that as a way to accelerate our internal progress, right? Because we’re building models, we’re building hardware. So if a frontier model can give us more data that way, like generate really tailored data for ourselves, that’s extremely useful for us. And another thing we’ve been thinking about is instead of using these models to generate 3D environments, we have also used our own world model, right, the world action model I was telling you about, because the model not only generates action, it can also generate videos. We— The model can also generate its own training data. So that becomes a self-fulfilling flywheel that I think will also be powerful. So overall, I’m just really excited about this direction because I think even though we have scaled to a million hours of data, we are deploying a bunch of robots. The robots are act— already starting to do very cool things. We’re just at the beginning, just the tip of iceberg for what is there to come. Yeah.

On-Device Inference, Latency, and Model Distillation

Vibhu [00:16:51]: For those on the research side, we recently had Runway on the podcast. One of their things is robotic world models. They’re, shifting a little towards there, so … nudge people to look at it. One of the questions I have about you making your own models is what is the constraints you operate in? So if you draw a parallel to language models, you see GPT-OSS 20B is very fit for your average MacBook. Then there’s a larger one that’s fit for a single GPU. Do you work in the constraints of robot, of model on on device, in cloud, a hybrid? And then how does that affect your scaling law of data? A million hours is great, but like do we just scale up parameters, data?

Jason Ma [00:17:38]: No, that’s a great question. So I think what’s really different about robots versus language model and other applications is that for coding agents, language model, it’s okay if it’s— it has latency, right? Well today, whenever I ask ChatGPT, it takes like couple of seconds before it spits out anything. But for robots the inference latency has to be very short, right? The robot needs to act. Like, it can’t just hold a knife and not do anything for three seconds. That’s gonna freak people out, right?

Vibhu [00:18:07]: Yeah, we saw someone put their real hand under a knife.

Jason Ma [00:18:10]: Our model’s, pretty smart, so that’s okay. But in general if you think about very dynamic tasks, like for example, you want your robot to hold a I don’t know, a tennis racket and juggle the ball, right? It’s just very dynamic, real reactive. You can’t just like. The physics, gravity doesn’t wait for you, right? So for that reason, robotics model needs to be able to inference very fast. And the fastest way to do that is by serving on-prem, like on device, right? Because if you call a model-

Swyx [00:18:39]: Network latency.

Jason Ma [00:18:40]: There’s no guarantee-

Vibhu [00:18:41]: Self-drive a car.

Jason Ma [00:18:42]: There’s no guarantee, right? Like, so for that reason, the model that gets deployed eventually has to be of reasonable size that can be served locally. But that doesn’t mean that the initial model you train also has to be small, right? Like in language model-

Vibhu [00:18:57]: You can distill.

Jason Ma [00:18:58]: People do distillation, right? Like, the best way to make a small model very smart is by doing distillation, right? You train the biggest model, then you distill, right? So I think in robotics, we’re getting there, but I think that’s what people are going to converge on. Yeah.

Vibhu [00:19:14]: How about inference? You have your own inference stack or. A lot of this is architecture. You’ve pulled diffusion and whatnot. I’m guessing you can pull off of something. Anything you wanna share?

Jason Ma [00:19:26]: So we build our own inference stack for whatever model we deploy, because that’s— like I was saying latency is so important. So owning the whole stack is extremely important for robotics, right? You want to optimize the inference speed for the kind of model you’re running, and there’s a lot of work in there. So that’s what we do.

Building the Full Robotics Stack: Sensors and Actuators

Swyx [00:19:45]: You’re really full stack, like- Everything from the actuators all the way to in the industrial design ID, which I- Didn’t really know as a acronym. Obviously to the models and data as well.

Vibhu [00:19:57]: I’m curious are these your own actuators, or are they. How much of this is your own? Like, of course, the design, it’s not like humanoid- Actuators, like how far down the robotics stack?

Jason Ma [00:20:08]: So the way we think about, what we need to own versus what we can buy is like if the Input-output and of a component is very clear, and like there’s already extremely good mature solution, then there’s no point of building in-house. For example, on this robot, you see the LiDARs, right? So those, black circle, those are LiDARs to help the robot detect obstacle, navigate surroundings. So that’s a very mature technology that people have been using for now decades. It’s on self-driving car. So we don’t need to build LiDAR from scratch, right? Same goes with some cameras. But for anything that we think is extremely crucial for the whole system’s performance at the end, our vision is to be able to own that and build that ourselves. So one thing is, you mentioned actuator. So that’s something we’re working on and we’re going to own ourselves. The reason is that like, the quality and how you design the actuator really matters for how well you can model the robot, right? Actuator is what drives the robot. To be able to model your robot very well in simulation or to do any kind of modeling, you need to select actuators very good that can actually be easily modelable. So that’s what we care about eventually. So we’re gonna own that. Same thing with the robot’s proportion. This robot’s proportion is we measure it against real people. Why do we do that? ‘Cause we’re pre-training on human data.

Swyx [00:21:36]: Oh, yeah.

Jason Ma [00:21:36]: Best transfer, right? So there’s a lot of these kind of decision that goes into hardware that’s very much tied to like. Because eventually, it’s hardware that carries intelligence. Intelligence is the extremely important thing. We want to make sure the hardware serves for that.

Swyx [00:21:53]: I would just highlight, I don’t know if this is relevant to you, but one of the most interesting robotics discussions I’ve had is someone who was very stressed out about actuators and specifically magnets, ‘cause apparently we don’t make magnets in the US.

Manufacturing, Rare-Earth Magnets, and Global Supply Chains

Jason Ma [00:22:06]: No, we don’t.

Swyx [00:22:07]: And there’s only two countries that make magnets.

Jason Ma [00:22:09]: Rare earth material.

Swyx [00:22:10]: Rare earth, right? I don’t know what the. Wh-what’s the chemi— what’s the material? I don’t know what the thing is, but- so yeah, if anyone’s interested in that’s, apparently a bottleneck that we have.

Jason Ma [00:22:20]: No, not bottleneck.

Swyx [00:22:21]: It’s a bottleneck. So what do we do?

Jason Ma [00:22:23]: It’s a good question, right? I think, there’s a lot of exciting discussions on bringing manufacturing back to the US, but for robots especially, that’s actually much harder than what it seems. And I would say even now, there isn’t a company in the US that’s truly building everything from scratch in the US, right? Like, even when they say they assemble robot, they build whatever they’re actually doing in the US, they’re still sourcing their raw materials from other places in the world. And I think that’s something to really be aware of as a robot builder, right? Like, I think, like as a full-stack robotics company, we realized just how long the supply chain is and how long the whole problem is. It’s actually extremely hard. And I think these days, a lot of the discussion centers around a robot foundation model data, but I think hardware and just how hard it is to build good hardware is very under-emphasized. But I think that’s gonna be extremely important.

Swyx [00:23:27]: This is a platform, right?

Why Dyna Co-Designs Its Hardware and Models

Jason Ma [00:23:29]: Platform, yeah.

Swyx [00:23:30]: There are some companies that choose to be, let’s call it platform-less. They only do the model, and they will say, “We deploy on any platform.” We deploy on Optimus.

Vibhu [00:23:37]: Off-the-shelf robotics.

Swyx [00:23:38]: And so, it’s a choice, right? Like, obviously, you own everything, so you can do that. Do you see yourself building more in the US, or do you see building your own global supply chain? What’s the direction that is likely?

Jason Ma [00:23:52]: So we are currently exploring many different options, right? Because with global supply chain, then there is some, geopolitical things, right?

Swyx [00:24:02]: Shipping.

Jason Ma [00:24:02]: Shipping. Also shipping delays and all that stuff.

Vibhu [00:24:04]: Literal ships for robots.

Jason Ma [00:24:06]: And in the US, it’s just that going from zero to one will take some time, right? So we’re trying to think about a solution that can both allow us to build more in the US while leveraging global supply chain, right? But I do think, going back to your previous question on some companies choose to just build model, we build everything, I think it’s very important to get this right. I feel that for robotics, the hard thing is. Like building the whole system. What are the necessary things to build a whole system? Like, previously, I was at DeepMind, I was at NVIDIA, primarily just building models, and I saw the upper bound of the system when you don’t control the hardware, when you don’t get to optimize that right? Because if you think about the value creation for robots, eventually robots are here to deliver a service to solve some physical tasks. So if you wanna solve physical tasks, you wanna be able to optimize everything, and that’s important.

The Semi-Humanoid Robot: Wheels, Torso Movement, and Dexterity

Vibhu [00:25:02]: We have the robot. You gotta tell us about the design. Your own custom platform. What do you call this?

Jason Ma [00:25:11]: What do we call it?

Vibhu [00:25:12]: Yeah, so it’s not humanoid. It’s very smooth, but what’s the what’s the rotating arm?

Jason Ma [00:25:16]: Yeah, so we think of this as generally like a semi-humanoid. So basically, the upper body is, like a humanoid. It’s got, of course, shoulder, got arms. It also got four degrees of freedom on the waist. So you can basically do all the upper body motion that a human can do. But it’s a semi-humanoid because the lower body is wheels. All right, so why do we choose wheels? Because I think there is a difference. There’s, optimal practicality versus performance trade-off. I think with legs, the energy con— the energy consumption would be a lot higher. It would be a lot more unstable. Safety will be extremely hard to guarantee. So right now, for legged robots, there’s actually no guarantee that the robot will stay upright, balanced the entire time. I’m sure you guys have seen a lot of those, robot dancing videos. Like the robot dancing thing.

Swyx [00:26:08]: Seeing dancing, yeah. They all, they fall over very easily.

Vibhu [00:26:10]: No, but you also see the demos that you can tell are not production, where you can box the robot and you can push it over and it’s very good. But that robot is just good at balance, right? Like- there’s a.

Swyx [00:26:21]: I think the more interesting thing that you guys have innovated. So we have, one of the prototypes, which is the Mewtwo back here.

Jason Ma [00:26:29]: You can call it that.

Swyx [00:26:30]: This is just like an up-down, axis, and you’ve chosen to then move towards something like this which, has the ability to lean. And we’re gonna show the laundry demo where you have to lean into machines. Right? That’s another degree of freedom that I don’t even have the words for it. It’s not pitch, it’s not yaw. It’s something else.

Jason Ma [00:26:47]: So yeah, how did we think about that right? Like,. So actually, it’s very used for exercise. Like, I think sometimes, a couple years ago when I was exercising, I accidentally like, pulled a muscle. Right? I’m sure you guys have that experience too, right? Then the next few days you just couldn’t move, or like you pull it where you slept in the wrong position, so your neck is very stiff. We have all experienced that. And then whenever you experience that you kind of realize, wow, so many muscles, so many degrees of freedom that we use in our daily life that we take for granted. But when it’s disabled it’s very awkward for you to do some tasks that you feel are very natural, right? So that’s also how we think about when we design robots. When you have a— this kind of design, which is this purely elevator, could only go up and down, we realize a lot of the human torso movement, like we use our waist, back a lot for day-to-day physical activity, right? Even if I wanna get this bottle I’m not just moving my arms. It’s actually very strange for me if I have to intentionally think about fixing my upper body. I would always do this right? So there’s just a lot of subconscious way that humans move their body that I think will be equally useful if robot has a degree of freedom. And if you further think about, okay, we want to learn from human data. Like, humans are moving their torso all the time. You wanna be able to ingest that into the model to control the robot. But if you have a form factor that doesn’t allow for that then the transfer becomes less and the robot becomes less dexterous and it’s just more awkward.

Transferring Human Pretraining to New Robot Hardware

Vibhu [00:28:22]: What’s the data labeling side of this? Like, I— you’ve seen people,. Is this the last post-training fine-tuning side of actually the hour of using the robot tele-operated or remote-controlled mapping motion? Or does some of it generalize over?

Jason Ma [00:28:41]: When you say labeling, you mean?

Vibhu [00:28:43]: Like, taking the model you have that predicts state to mapping it to your specific hardware.

Jason Ma [00:28:50]: Oh, yeah. It generalizes very well. So for all the demos that I have shown you on the new robot, ‘cause the new robot, we, it got ready not so long ago. So we quite literally do not have a lot of data compared to the stationary arms we showed you, which had much more data, right? But a new robot, the model has not seen this in pre-training, in some of our pre-trained model. It’s able to learn the task that you saw very quickly. And I credit a lot of that to like, even before the hardware is ready, we already committed to pre-training on humans because we knew at some point our robots will be like human. You can get ahead on the training side while hardware is getting ready. And I think now we’re seeing that the transfer is actually there and is actually very helpful.

Evaluating New Skills and Learning from Real-World Feedback

Vibhu [00:29:35]: Talk about the eval side of this. So you can have predictions, you can have it do stuff, and every now and then we see it miss stuff. You test on very niche things as well. It’s like filling a pill bottle. How long does that process take? How do you think about the aspect of actually deploying on a new task, evaling it, iterating? What does that look like?

Jason Ma [00:29:56]: Yeah, so if we wanna learn a new task, we basically do a little bit of robot data collection. Then the model, these-

Swyx [00:30:04]: What is a little bit like?

Jason Ma [00:30:05]: Yeah, like couple hours. Yeah, we’re— In some of the tasks, like just tens of minutes, right? Yeah. And then-

Swyx [00:30:11]: The chopping one was 45 minutes.

Vibhu [00:30:13]: The open the fridge was on-

Swyx [00:30:14]: And that’s an interview— That’s like an onboarding task for one of your engineers.

Jason Ma [00:30:18]: So,.

Vibhu [00:30:20]: Fun thing if you wanna work here, you know. Find the task, have the robot do it.

Jason Ma [00:30:24]: Pitch me a new task and I’ll come to the office, collect some data, train a model and see what happens.

Vibhu [00:30:29]: Tie your shoes.

Jason Ma [00:30:31]: Truly, it’s tying. Yeah, that would be cool. But yeah. So, you know. A little, a small amount of data collected, train a model, then we just deploy the model in the real world. And then we look at the performance, then-

Vibhu [00:30:41]: To fill in collected means you are— you have a person actually operating-

Jason Ma [00:30:46]: Operating.

Vibhu [00:30:48]: Doing the task for a bit. And then you train? And it does okay, you improve. How do you measure?

Jason Ma [00:30:56]: So when you’re training the model, there’s, certain metrics you can look at right? Just like training a model, it converges nicely. Now, after that for robotics today, we just run the model in the real world, right, on the same setup that the data is collected to see how the model does. Sometimes the model just works great out of the box, right? You don’t really need to do much more. But when the model’s maybe like seventy percent, eighty percent there, then you see how the model behaves. You figure out what it’s not good at. Then you can collect a bit more data, then iterate, right? I think of this as the robotics version of reinforcement learning from human feedback, right? The robot rolls out. You can score it good, bad S- somewhat good. And then that data, the robot’s own experience can be used to train the model. And you can also just collect more data yourself to train a model, right? Eventually, our goal is like, this robot is gonna be our platform. Anyone should be able to train tasks on it.

Long-Horizon Tasks and the Bottlenecks to General Robotics

Vibhu [00:31:59]: I wanna get a bit of a prediction and timeline to where that goes. Right now, the tasks are, somewhat simple, like- Take laundry out. It’s a towel. You can’t crush it. Chop some food. There’s also the side of like, okay, set up this filming studio, right? Move stuff around, bring a table, set up lights. I think-

Swyx [00:32:21]: Manage my data center. And also, ref— like, recharge another robot.

Vibhu [00:32:29]: When do we see the factory or what do we need to scale up? Is it model? Is it everything? It seems like your hardware is set for that but where do we see it go to a point where I can go from instead of just cut one thing to, “Okay, here’s a recipe.”

Jason Ma [00:32:49]: I would say to get to truly general-purpose robots like basically a robot or a model that does anything, like-

Vibhu [00:32:58]: Not that. Like, okay, so to keep it somewhat grounded on a technical level, like- when do we start to see us pushing that feedback, right? So-

Jason Ma [00:33:06]: I see.

Vibhu [00:33:06]: Instead of very simple chopping- Maybe a five to ten-minute sequence, right? Where now the eval step, it can break at many more points. Instead of just fill one pill bottle- Ten different pills all in a sequence, right? What’s limiting us? What are we pushing on?

Jason Ma [00:33:20]: Yeah, I think. Yeah, I think a couple of things. One is that the laundry demo that you have seen, yeah, it’s gonna be one hour to finish a whole loop, right? So the longest duration the robot can do a full job is already in the hour range. Right? But to be able to do that for any task, I think the bottleneck is coming back to data and hardware, right? So for some of the tasks, for example, setting up the space, you might need more than a the parallel jaw gripper, right? You might need something more dexterous, like a hand, like a human hand, right? But then if you wanna train a model on the hand, then you’re bottlenecked by the hardware stability. Like, hand— Today, the best hands, dexterous hands are still not, durable enough. They overheat pretty quickly, right? So that becomes a bottleneck, right? Like, you— We— you quite literally have to put a ice bag-

Swyx [00:34:15]: Ice. To cool.

Jason Ma [00:34:16]: Human.

Vibhu [00:34:16]: We’re not just auto-researching, make better model we have. We have hardware limitations.

Jason Ma [00:34:20]: Yeah, so hardware limitation is a big thing. But then if hardware is limited, then that will slow down everything, right? Because to know that you’re making progress on any task, you eventually have to eval on the hardware, right? So then without the eval being very spontaneous, very fast, that also limits your research progress, right? Because then how do we collect the right data, right? I think, how do we collect the right data to do extremely complex, like setting up this studio, right? Because you can imagine, like today, most of the data is vision-based, right? It’s like video streams and some, encoding of your hand motion where the robot’s, joints. But Humans have a lot more sensors. We hear things. We feel things.

Swyx [00:35:05]: Smell.

Jason Ma [00:35:06]: Yeah, you smell things, right? So we subconsciously use all of those to help us be so good at manipulating the physical world. But today if we’re not capturing that at all, then even when we scale to millions of hours, you might just be missing critical dimensions of the task, so you’ll never be able to solve the task, right? So I think those are what I consider what’s dragging on the timeline. But for tasks or for workflows that you can collect data today to high fidelity, then I don’t ne— I don’t necessarily think there is a fundamental bottleneck, right? Just like what we are demonstrating, to the public this week. We have a physical agent that can do the entire laundry workflow, right, as a human would, right? Because for that task, we don’t really have like, oh, you actually need more than a parallel jaw gripper. You actually need some sensing modality that’s just not available today, right? So I do think the biggest bottleneck To have a much more general thing is still hardware and data.

Real-World Deployments: Best Western and Din Tai Fung

Swyx [00:36:06]: I want to take this chance to talk— let you brag about your customers and partners. You have some announced, relationships. We can start with Best Western and then maybe some of the others that you’re, that you’re proud of. Cool stories, anything like that.

Jason Ma [00:36:20]: So we have a deployment blog post, on our website that goes into much more detail. I really recommend watching the video because I think the video is-

Swyx [00:36:29]: Which one? The Best Western one?

Jason Ma [00:36:30]: No, we have one video for all the customers we’re working with. It’s a really nice video. It shows what we’re doing. And yeah, I think it’s great. But yeah, we have customers, Best Western. That’s in the hotel space. We fold a lot of laundry towels for them. Every hotel needs it, every gym needs it, every fitness center needs it. -

Swyx [00:36:51]: What’s the deployment so far? How— Like, where can I go to— Which Best Western do I go to see it?

Jason Ma [00:36:56]: So if you wanna go somewhere, I actually recommend going to our other deployment, Din Tai Fung. Have you guys been to-

Vibhu [00:37:04]: Very thin wrapper.

Swyx [00:37:05]: We should have filmed there.

Jason Ma [00:37:09]: Just filmed there.

Vibhu [00:37:10]: The B-roll.

Jason Ma [00:37:11]: But in Din Tai Fung if you go to Din Tai Fung or other somewhat upscale restaurant, there’s napkins on your table, right? So our robots today fold thousands of napkins for many different locations, many different Din Tai Fung locations in the United States. So that’s actually our biggest scaled customer, right? So I wanna make a distinction here. Today, there’s a lot of robotics companies that’s talking about deployments, talking about pilots. But what’s really hard is not putting your robot in one location or doing a pilot. What’s really hard is that the customers believe or validate that your robot creates enough value that they wanna deploy your robots across many of their sites. So that’s what’s happened for us with Din Tai Fung, which, to the best of my knowledge, hasn’t really happened for other companies doing similar general purpose robots. So at Din Tai Fung, our robots fold thousands of napkins per day. It’s — The quality standard is extremely high, ‘cause they want like, okay, the second fold or the seam of the napkin is only like a couple centimeters below the I don’t even know how you call it, the pocket, so you can stick the chopsticks in. So it’s actually very fine-grained, high-precision task. I can show you.

Vibhu [00:38:26]: At some level, that’s what robotics are good for robots are good for too, right? Humans are not the most consistent- Especially in small details, whereas a robot can somewhat do it perfectly. It’s interesting that’s the example you bring for general purpose robotics versus, at some level, I think people would assume- you can have a specialized robot that will fold napkins perfectly to the exact, millimeter as opposed to something like this but.

Jason Ma [00:38:53]: Yeah, I’m sure, maybe you can, right? But the hard thing about manipulating deformable object, right, like napkin, towels, everything’s- Deformable. There’s just infinite- infinitely many possible configuration. So that’s really hard for a specialized robot, right? And what we find is that by developing a general purpose model in a robot and adapting it to these use cases, you can actually get to a production quality faster. And what’s really nice is that now that Din Tai Fung has scaled, has signed a much bigger contract with us, they not only want our robots to fold napkins. Now they want our robots cut. That’s why we had the- Example, to cut food to the-

Swyx [00:39:31]: Is that actually the second use case?

Jason Ma [00:39:33]: That’s one of the use case they are thinking. Din Tai Fung needs to cut so many vegetables per day.

Swyx [00:39:38]: They need to do the dumplings.

Jason Ma [00:39:39]: Exactly. To the fold. That’s the final boss. That’s-

Vibhu [00:39:41]: That’s the final one.

Swyx [00:39:43]: I think you can. I think you have the dexterity and the precision to do this.

Vibhu [00:39:46]: You have a little robot.

Swyx [00:39:47]: But no, don’t break the dumplings.

Vibhu [00:39:50]: That’s the obvious benefit is you can adapt them anymore.

Generalist Robotics, Speed, and Consistency at Scale

Vibhu [00:39:54]: I think the other thing you really touch on is the speed, right? To make a custom napkin folding with the exact ply, whatever- you have to make the machine. You can’t just Amazon this. But this can do it pretty fast, right?

Jason Ma [00:40:07]: Yeah, this can do it ex— very fast. And now it’s actually al— it’s at human parity. And this robot doesn’t get tired, doesn’t need to you know. You can. And I think I really like the fact that you mentioned consistency, right? Like, one issue with Din Tai Fung is like, their LA location with New York location, Seattle location, the they fold napkins slightly differently, but from their perspective, they want consistency. But that’s exactly what these models are good at, ‘cause models are, just, it’s a set of weights. You can just put the weights in the different robots across different-

Vibhu [00:40:43]: You can OTA update them as well.

Jason Ma [00:40:45]: Yeah, you can just update all of them at once. It just. Once it works, I think it’s truly, like- Incredible.

Vibhu [00:40:52]: Final boss is Cheesecake Factory. If the story there, they’re from scratch. Everything is very consistent, down to the measured gram. Every sauce is made. Like, anywhere you go in the world, you’ll have the exact same thing. If you don’t know, you don’t know. But that’s the that’s the final.

The Bitter Lesson of Robotics: Scale the Right Data

Swyx [00:41:07]: The term that comes to mind is kind of like bitter lesson for robotics, right? You, instead of the specialized robots, which is a lot of industrial engineering, you build a generalized platform that can- Do a last-mile fine-tuning and-

Vibhu [00:41:18]: That’s right.

Swyx [00:41:19]: And do all those things.

Vibhu [00:41:20]: Give the expert add-on outside of your. What is more bitter lesson robotics?

Jason Ma [00:41:25]: Yeah, I think there’s several bitter lessons for robotics. One is that robotics is going to scale just like any other domains that machine learning has solved, right? So what does that mean? That means having as much data as possible. But in robotics, the one thing that’s a lot harder is what is the right data is much harder to figure out than other domains, right? Language is text. In video games, just video game data. But for robotics what sensors you choose determine what kind of data you have, right? But if you have the right set of sensors and just scale that data, I think you’ll get predictable return on model performance. So in that way, robotics isn’t so different. So that’s the bitter lesson. As a corollary of that is that like there’s a lot of research in robotics that’s coming out with very specific priors. Like, for example, we need to make the robot see objects as they are, draw some boxes. That will give robot better perception, right? But I think the bitter lesson here is that just keep the algorithm simpler and scale the right kind of data. Then generalization, better behavior will emerge. So that’s not so different. And the second. I think that’s the main bitter lesson, which is at the end of the day, scaling will work, but in robotics, to scale and how to scale is just much harder ‘cause there is no internet. You have to come up with your hardware, come up with the sensors, get all of those right, and then scale. Then I think,.

Humanoids, Practical Iteration, and the Path to Generality

Vibhu [00:42:53]: There’s the argument that the world is very made for humans. It should be a human, right? As hard of a hardware problem that is, think ten, twenty, a hundred, two hundred years down the line, can we make a really efficient human? But you have to be practical with what we have now, right?

Jason Ma [00:43:12]: So the way I think about it is that yes, eventually the robot should. If we wanna build a general-purpose robot, right, like a machine that can be adapted to as many use cases as possible, I think human form factor will be the best one, just because that’s what we have been doing, for ages, right? But I think the best way to actually build a humanoid robot is to take an iterative approach. Because robotic is so hard, the whole problem, the whole stack is so long, it’s really hard to know a priori what’s the right decision to make, right? And because hardware decision is much more expensive and much more costly, just ‘cause hardware iteration is a lot slower, you actually want to take an incremental iterative approach, not going all the way to the final form factor. Because the chances are the thing you build actually doesn’t last. It’s not good enough. So you just end up scratching it anyway, you know. Like, it’s public news Optimus, already gone through many iterations, and they had to scratch their designs, right? And I think by iterating with a semi-humanoid form factor and actually putting them out to real customers or in getting models really deployed, you get that feedback loop that you wouldn’t get if you’re just trying to build like, warehouse humanoid and you spend like five, ten years building it. I think the iterative approach will actually get us closer to the final robot faster.

Vibhu [00:44:33]: And I will say, for those listening and not watching, it’s probably the smoothest robot we’ve seen. Like-

Continuous Action Spaces and the Importance of Context

Vibhu [00:44:38]: … every humanoid you’ve seen is jerky and even humans don’t walk as smooth as this thing.

Swyx [00:44:43]: Because, yeah, the model is. Can move every single motor at once.

Vibhu [00:44:48]: In more than two cohort, yeah.

Jason Ma [00:44:50]: Yeah, and that’s actually a very hard problem, because if you just think about the model, it’s outputting action. There’s the action, dimension, right? The bigger the dimension is, the harder to fit, the model to the data. But here, the action space is quite large, but we did very good job to make sure the model fits well and gives you a smooth-

Swyx [00:45:10]: Can you bound the order of magnitude of the action space, or is that-

Jason Ma [00:45:14]: Yeah, so it’s on the order of, about a hundred action dimensions.

Swyx [00:45:19]: Yeah, and these are all continuou- continuous, right?

Jason Ma [00:45:21]: Continuous, right. So the model is just basically predicting at very high frequency, like tens, like at least fifty-dimensional vector, like ten, twenty times a second to get that continuous motion.

Swyx [00:45:34]: And potentially predicting one to two seconds out, so that is another hundred frames or whatever.

Jason Ma [00:45:39]: Of course.

Vibhu [00:45:40]: There’s a bit of a history, similar to context of previous state, I guess.

Jason Ma [00:45:45]: So context is a very important thing in robotics, right? Because otherwise, if you’re just making decision based on the current observation, sometimes it’s really hard to tell what to do next, right? For example if you think about opening a microwave, like in the middle part, it’s unclear whether or not you’re-

Vibhu [00:46:01]: Closing

Jason Ma [00:46:01]: … closing or trying to open it, right? So context is also really important.

Swyx [00:46:05]: And I have a screenshot of the fridge demo where it is, there’s agentic planning and tool holding and all those things. So we can show that for listeners.

Vibhu [00:46:13]: I think that’s an interesting side. After the model-

Agentic Harnesses and Planning Physical Workflows

Vibhu [00:46:16]: … how do you wrap it around into agentic harness? Like, what is-

Swyx [00:46:22]: Claude Code

Vibhu [00:46:22]: … Claude Code deployment? Claude Code UI. You have a lot of cool UIs, but what goes into wrapping model output towards final thing? Is this custom per use case? Is it you have your own SDK?

Jason Ma [00:46:37]: So, eventually we will have our own SDK, but the way we think about is we want a general harness layer for robotics foundation model. Today, that is our own model, right? The way I think about is for any Real task, long task in the physical world, there is a lot of a high-level decision-making you have to do, right? It’s not just like robot, only reasoning about the physical world in the action, continuous trajectories, right? Because that’s a very low level, the thing you’re actually executing in the physical world, but that’s not the intent. But like if. When we are doing decision-making, we think about steps of a recipe. We think about, oh I finished this I need to do that. While I’m waiting for this part to finish, I need to go somewhere else to do something else. There’s a lot of high-level cognition, decision-making that’s happening. And I think for robots to handle even, the laundry task we are showing you, it has to have that capability. Because, it needs to reason that okay, the washer is running, so I should go do something else while the washer’s finishing. But now the washer is done, I should take, I should take the towels from the washer and put it in dryer immediately, right? Because if not, then that becomes a bottleneck in the whole workflow, right? So there is a lot of reasoning that needs to happen for robotics. And I think in theory, you could just, try to learn that with a low-level model that’s outputting actions, but it’ll be extremely inefficient. So that is why having a hierarchical system, like putting a harness, like an agentic system around our world action model is what allow the model to compose different capabilities, different skills to solve a much longer horizon task. It’s very much like in language model, right? Like two years ago, three years ago, the model was only good at autocomplete. Like, humans have to tap to get the model to keep spitting out code. But today, just an agentic system just carries the end-to-end task. And that’s. I, and I think that’s important because I think today in robotics, people are still thinking about autonomy at the very task level like one minute, thirty-second task. But really, the value is about tasks or workflows that go on unsupervised for hours. And that’s what we’re building towards. And that’s why I think thinking about. The robotics problem, not just as a modeling problem, but rather as a system problem, you need good agent, good harness engineer, good model, good hardware, everything. It’s like how we think general-purpose robots ought to be solved.

How Coding Agents Accelerate Robotics Research

Vibhu [00:49:07]: Do you guys see benefits from language model coding tools getting better? So new Opus 5.5, new Astra. Do you have language models in your agent harness? How does AGI being developed in language model space affect you guys?

Jason Ma [00:49:24]: No. Yeah, no, absolutely. Like, it affects us very deeply in many ways. One is that we are extremely AI-native company, right? Like, coding agents, all of that is a very fundamental part of our work, and we’re very welcoming of that. Like, whenever there’s a new model, we’re the first one to test it, right? Like, we being. Like, when GPT-6 came out, we had a company-wide hackathon the same week, and the thing that our engineers built was incredible. And many of them are— Many of the features we built is already in main, in production, right? So we really like using coding agents to accelerate our work.

Swyx [00:50:00]: Just for the people interested, anything surprising in GPT-6 that you found?

Jason Ma [00:50:05]: … it’s about the ability for it to construct virtual environments, right? Turns dissemination. We talked that twenty minutes ago. I think that’s really exciting.

Vibhu [00:50:17]: It’s interesting. There were so many CAD, Zillow to 3D-ish.

Swyx [00:50:21]: He lost the hackathon to the CAD, -

Vibhu [00:50:24]: I won the. I did the CAD b******t. Other way.

Swyx [00:50:27]: No. Oh.

Vibhu [00:50:28]: That was. I won a GPT hackathon just for fun. I was like, “I know nothing CAD.” Might as well play around with it ‘cause every demo is doing it. I’m like, “It’s gonna be garbage.” And I’m like, “Well, it’s kinda grounded, it’s kinda good.” But it’s just hackathon stuff.

Deployment Milestones and Competing Robotics Strategies

Vibhu [00:50:45]:.

Swyx [00:50:45]: Cool. I wa— I was gonna round out, some of the use cases and then go into a little bit of company culture and your personal background.

Vibhu [00:50:52]: I have two technical questions. One is, are there any goals you wanna set for yourself, like six months, one year, two year timeline that you see robots should be able to do, fill in the blank, that you’re kinda working towards?

Jason Ma [00:51:09]: Yeah, in the six months to one year frame, our goal is definitely scaling up our robots and our model and truly getting them to deploy-

Vibhu [00:51:18]: Like a task.

Jason Ma [00:51:18]: … completely different.

Vibhu [00:51:19]: Like, right now it’s cutting or opening a fridge. Is there. Like, do you set yourself to by this timeline, we should, to stay on track, be able to do whatever?

Jason Ma [00:51:29]: I think, in terms of deployments, we’re exploring many different industries. There isn’t a particular task we think we have to nail, right? Because our goal is to build a general model, a general platform, and we think of each different task as validation of our platform as a way, as a wedge for us to figure out how to build everything better. So today, we are doing a lot of that in laundry, in restaurants. We have already deployed tens of robots, but our goal next year is overall just deploying hundreds of robots.

Vibhu [00:52:02]: My part two on the last technical question is, how do you see the rest of the industry? So fill in for people, what other approaches are people taking to world models? Why are they wrong? What is different but good about their approaches as well? Just what is the landscape of robotic world models in this space and why— how have you come to your current?

Jason Ma [00:52:25]: No, that’s a great question. So I think robotics is the most exciting space today, and there are just so many, amazing companies that I truly respect. They’ve been doing great frontier work like us, and it’s very exciting. And I think what we are seeing today is that there’s still many different approaches to robotics. Like, we believe that hardware needs to be built in-house. We need to own the whole, chain. We believe that scaling human data is extremely important. But there are companies that’s betting on just scaling robots’ data or scaling some sort of intermediate data collection devices, right? All of them are in play, and I think it’s different approaches. We’ll just see what happens. I think right now, it’s still a bit early to tell, but we feel pretty good about the way we’re building things just because we as a company, culture-wise, mission-wise, we’re not here to just build a model. We’re here to truly get robots deployed. So we deploy as much as possible, work on the hard problems. So we identify problems much earlier on. So then I think that really helps us identify what the true bottlenecks are and then solve them. And I think for folks, for people who are only working on one part of the problem, it’s kind of like fog of war. You don’t really see the whole picture, and you might be building something that’s extremely cool technologically, but may not be what people need or want at the end, right? I think there’s a whole graveyard of robotics company, and I think all of them had very shiny, very cool robots, but just figuring out wh…

From the Bitter Lesson of AI scaling to the unsolved mysteries of protein folding, Google DeepMind’s Pushmeet Kohli and Biohub’s Sal Candido are rethinking what it takes to build AI that truly understands biology. In this special panel moderated by Brandon Anderson, they explore why AlphaFold’s breakthrough was only the beginning, why scaling compute and data alone won’t solve biology, and how the next generation of AI models could transform our understanding of proteins, cells, and human disease.

We go deep on the future of AI-driven biology: finding scaling laws in biological data, the tradeoffs between scientific intuition and general-purpose architectures, why protein structure prediction is far from solved, and what it would take to build predictive models of living systems. Pushmeet reflects on the lessons behind AlphaFold, the limits of human interpretability, and whether future frontier models could understand other AI systems better than we can. Sal explains why protein language models may already contain scientific knowledge we haven’t unlocked, how biological modeling must move beyond individual proteins, and why achieving Biohub’s mission to cure all disease requires thinking in terms of 10x breakthroughs rather than incremental improvements.

We discuss:

* The Bitter Lesson for biology: why scaling compute and data isn’t enough

* Why finding the right scaling law matters more than blindly increasing model size

* How low-quality metagenomic data can improve protein language models

* Why AI researchers optimize for available data instead of the most important scientific problems

* Lessons from DeepMind on balancing modeling, data generation, and scientific expertise

* Why building a virtual cell requires fundamentally different datasets

* AlphaFold’s handcrafted architecture and the role of scientific intuition

* Why good data matters more than simply having more data

* Inductive biases, scaling laws, and the future of specialized AI architectures

* Why we aren’t in a post-Transformer world, but architectures are evolving

* Why AlphaFold didn’t actually solve all of protein folding

* Protein dynamics, disorder, and the limitations of static structure prediction

* How cryo-EM micrographs could unlock richer biological representations

* Moving from models of individual proteins to whole biological systems

* Feynman’s famous principle and why AI can now create things we don’t understand

* The hidden biological knowledge inside protein language models

* Why trustworthiness and uncertainty calibration matter more than full interpretability

* Whether frontier AI models could interpret other AI systems better than humans

* When AI could deliver 10x–100x acceleration in drug discovery

* Why curing all disease requires thinking about 10x breakthroughs instead of 10% improvements

Pushmeet Kohli — Google DeepMind

* X: https://x.com/pushmeet

* LinkedIn: https://www.linkedin.com/in/pushmeet-kohli-4838994/

Sal Candido — Biohub

* Biohub: https://biohub.org/team/salvatore-candido/

* X: https://x.com/salcandido

* LinkedIn: https://www.linkedin.com/in/salcandido/

Brandon Anderson — Moderator

* LinkedIn: https://www.linkedin.com/in/brandon--anderson

Timestamps

00:00:00 Introduction: The Bitter Lesson for Biological Data

00:01:00 Finding Scaling Laws and the Right Data for Biology

00:04:54 DeepMind’s Bitter Lesson: Solving Problems vs. Scaling Models

00:06:50 AlphaFold, Data Limitations, and Building the Virtual Cell

00:09:52 Handcrafted Architectures vs. Scaling Compute

00:11:53 Good Data, Inductive Bias, and Model Design

00:13:39 Beyond Transformers: The Future of AI Architectures

00:14:38 Why AlphaFold Hasn’t Solved Protein Folding

00:17:39 Protein Dynamics, Design, and Cryo-EM

00:19:09 From Individual Proteins to Whole Biological Systems

00:20:43 Feynman’s Principle: Creating Without Understanding

00:22:18 The Hidden Knowledge Inside Protein Language Models

00:23:48 AlphaFold, Trustworthiness, and Interpretability

00:25:56 Could AI Understand Other AI Models Better Than Humans?

00:27:43 When Will AI Revolutionize Drug Discovery?

00:29:47 Why Curing All Disease Requires 10x Breakthroughs

Transcript

Introduction: Is There a Bitter Lesson for Data?

Brandon Anderson [00:00:04]: Great to be here. What an exciting morning. So many cool announcements. I think the future of bioscience is being announced right here. This is the modeling session. We’re all modelers, so of course it’s natural for us to talk about data.

One of the things I like to think about when it comes to data is how to scale it properly. This has brought me to the question of the bitter lesson, but recast in the frame of data. The bitter lesson, for those of you who are not AI people, is the statement that methods that scale win eventually. If you can scale enough, it wins. So my question, starting with Sal, is: Is there a bitter lesson for data?

Sal Candido [00:01:00]: For sure. You obviously need the right data in order for it to work. One misconception of scaling laws is that scaling laws are everywhere and they always exist. A lot of the work is actually finding that scaling law. A lot of what we do is trying to figure out: What’s a situation where, if you put more compute into it, if you put more data into it, you get a better result out?

That’s a great situation because, once that happens, you can turn the crank. It becomes an engineering problem, which is something I like. That has to do with architecture, but it also very much has to do with data. If you don’t have data with the right information and statistics to solve the problem you want, you’re not going to get a model that has the capabilities and understanding you want. You can only really pull information from the data you have and use that to generalize beyond it.

Choosing the Right Data: Availability vs. Scientific Impact

Brandon Anderson [00:02:09]: When it comes to data collection, how do you think about which modalities are best? A related question: Do modelers tend to work on problems where the data is available, rather than the problems that best serve their goals or have the greatest impact on translational medicine?

Sal Candido [00:02:32]: For sure. I won’t speak for all modelers, but I’m lazy, so I’m going to work with what’s available. There’s a good side and a bad side to this.

The good side is that, when you’re doing conventional machine learning, you’re often looking for the most pristine, high-quality data examples you can find. But if you’re like me, you’re rooting around in the back room, looking through people’s junk to see what’s there. A concrete example is training a protein language model on metagenomic sequences, which are not the highest-quality data. In fact, much of that data, I can guarantee you, isn’t even a real, whole protein. And yet it makes the performance of the model go up for designing real proteins that work and understanding proteins that we know exist.

That’s the positive side. But it can lead you to a negative place where you say, “Let’s just scale up the data that we can generate easily.” I don’t think that’s necessarily the way to do it. That’s one of the things that’s exciting to me about what we’re talking about here today with the BBI. It’s going out and asking, “What is the data that we need to solve the problem?”

It’s about the right resources, but also the right community. One thing we try to do at Biohub is work in the open, work with the community, and move the whole community forward. That’s critical because, if you just have people building models, we’re going to lean toward the data that exists. If you just have people generating data, they’re going to lean toward the things that can be generated. But if you can work together as a community, every step of the way, in an open fashion, you can figure out what data you actually need. Then you can find that scaling law.

The Problem Comes First: Modeling, Data, and Expertise

Brandon Anderson [00:04:50]: Pushmeet, what do you think about the bitter lesson for data?

Pushmeet Kohli [00:04:54]: I was at DeepMind when Rich was with us and thinking about this idea of the bitter lesson. What I took from Rich’s original lecture at DeepMind wasn’t about the specific notion of whether data is useful or how we should think about machine learning. My take was that he was talking about something more conceptual.

Sometimes when we’re looking at problems, we think about solutions in a very religious way: “I’m a modeler,” or “I’m a data generation person.” I think that is the bitter lesson. If you approach a problem with that mindset, you might not succeed. The problem comes first, and you should be flexible in your solution space. You should try both things. You should understand the problem you’re trying to solve. If it requires modeling effort, do the modeling effort. If it requires collecting more data, then collect data.

What happened in machine learning at the time was people saying, “We’re machine learning researchers. The dataset is there. Here’s some training data, here’s some test data, and we’ll just optimize the model.” That is broken in the sense that, if your eventual goal is to solve the problem, you have to look at both aspects of what goes into the process.

And what goes into the process is not just data or modeling. It’s also expertise. With AlphaFold, for example, we looked at what was possible with existing datasets because we didn’t have the core expertise, or even the resources, to say, “Let’s augment the PDB by a significant order of magnitude.” The investment that organizations and scientists across the world had put into constructing that data was invaluable. So there, you had to focus on getting the biggest bang for your buck by investing in modeling.

But in other areas, say cell genomics, we took the same approach and asked, “What can you do with cell-by-gene data?” After a lot of work, it was very clear that the data was not there yet to pursue that grand ambition of building the virtual cell. This brings people together to focus on the actual challenges of advancing science rather than religiously following advances in data generation or modeling.

Brandon Anderson [00:08:23]: I really like that answer. What’s the actionable takeaway? Always define the problem first and then figure out which solution space to search over. But more broadly, how should the community think about this as we move toward the next generation of translational medicine?

Pushmeet Kohli [00:08:47]: The advice I give to anyone starting in this area is to think of yourself as a multidisciplinary person. Understand the problem first. Why are you working on it? What are you trying to achieve? Then think about what’s needed, whether that’s modeling, compute scaling, or data generation. Understanding the problem is extremely important, and of course you need to build your expertise.

There are constraints, too. Maybe there’s only a certain amount of data you can generate, or a certain model size you can afford to train. Understand those constraints, try to fail fast, and look at which approaches will be feasible in the long term to get you to the level of impact you’re aiming for.

Scientific Inductive Bias vs. Scaling: The Craft of Building Models

Brandon Anderson [00:09:52]: This leads right into my next question. When I look at the evolution of modeling, AlphaFold 2 was essentially a work of art, with a lot of carefully handcrafted features. There was very intentional thought put into every part of the solution. Some of Google’s or Alphabet’s work still stays in that space, but the general consensus seems to be moving toward more scalable, general strategies.

Do we still need artisanal, craft solutions for certain problems? Or are resources better spent focusing on scale first? With fixed resources and money, you can invest in compute, talent, or data. How should we think about that trade-off?

Pushmeet Kohli [00:10:48]: Let’s approach it from first principles. The art and craft of constructing a better model wasn’t accidental. We did a lot of experimentation, but there was a vision behind it. Scientific intuition from biophysics and biochemistry told us that amino acid residues are not just doing their own thing. They’re influenced by other residues. So let’s bake that in. If you’ve learned something from the scientific community, use that information. Give the model that unfair advantage.

That makes the model much more data-efficient because it doesn’t have to relearn everything. I also have to mention that curating good data is an art in itself. It’s not as if you can just say, “Put in more data.” If you replicate the same data, you’re not going anywhere. It’s not just about big data; it’s about good data and understanding the coverage of the data necessary to make progress. That’s a much more interesting and challenging problem in itself.

Brandon Anderson [00:12:20]: It looks like you have a thought, Sal. What do you think?

Sal Candido [00:12:24]: I very much agree. It depends on the problem to be solved. If you have a smaller amount of data, having more inductive bias in the model is going to help you. You actually need that to get results at smaller data scales. As you get more and more data, sometimes the model finds things you didn’t necessarily know about, and sometimes that inductive bias, if it wasn’t exactly correct, can hold you back. There’s a tipping point as things get better.

But I also object to your question a little because there’s a lot of craft in the scaling part of things as well. There’s a lot of algorithmic work that goes into taking models, training them on more data, putting more compute into them, and making them bigger. And that isn’t only from an infrastructure perspective or about making the models run inference faster. We’re seeing things go beyond standard transformers to more bespoke architectures. I don’t think we’re in a post-transformer world in any way, shape, or form, but we’re modifying those architectures to make them more fit for purpose and work better, even with internet-scale data.

As more data comes in, the challenge isn’t only curating that data or selecting the next batch of information the models need. It’s also asking, at every step and at every scale, what the right architecture is to get the most out of that information.

Why AlphaFold Didn’t Solve All of Protein Biology

Brandon Anderson [00:14:36]: If you think about the state of protein structure prediction right now, the news will say that protein structure prediction has been solved. But if you talk to my friends, they’ll say, “We have so much left to do here.” Function, dynamics, and design are still wide-open problems. It’s been about five years since AlphaFold 2 was announced, and progress has been made.

Pushmeet, do you think there’s another big leap coming? Are there blockers to solving these problems? Do we not have the right data, algorithms, or ideas? Or is it just a matter of time?

Pushmeet Kohli [00:15:28]: Science operates by isolating something and then making progress step by step. When people say the protein-folding problem has been solved, at a conceptual level there have been advances. But think about the narrative of proteins being the building blocks. Proteins aren’t blocks, and they don’t act as blocks. I say proteins are the building blocks all the time, but I don’t actually believe it.

Proteins are extremely complex. They’re disordered. Their shape might change depending on the context. John Jumper and I used to discuss this: What are we trying to solve? We don’t know the actual true ground state that proteins take, or the actual distribution of structures that proteins take. What we were trying to do was replicate a structure that somebody had obtained and deposited in the PDB. That’s what we did. And it just so happens that it’s useful.

But that doesn’t mean we’ve understood all of protein dynamics. At the top level, it’s easier to communicate that we’ve made progress, but the scientists in this crowd know how much remains to be done. There’s a lot to celebrate, but let’s not stop funding protein structure prediction and protein dynamics, because we’re just getting started.

Cryo-EM, Molecular Dynamics, and the Missing Information

Brandon Anderson [00:17:39]: What’s the biggest blocker? If you could wave a magic wand and say, “We have more of this,” and that would accelerate function, dynamics, or design, what would you bring into existence?

Pushmeet Kohli [00:17:54]: My background is quite eclectic. I started as a security researcher, then went into computer vision, Bayesian theory, discriminative machine learning, deep learning, AI for coding, and finally science. So when you ask me that question, the computer vision researcher in me gets very excited about cryo-EM micrographs.

I thought, “What is this PDB data? I should be working at the source. I should be looking at cryo-EM micrographs. I don’t want those structures. They must be missing out on all the data. I should just operate directly on cryo-EM micrographs.”

Getting models that can scale at that level, with the right amount of data, and extract the dynamics and distributional information captured there would be amazing. I tried it, but it requires more work.

Brandon Anderson [00:18:57]: There’s still work to do, but you believe that’s a route that could give you that information?

Pushmeet Kohli [00:19:00]: Yeah. I think at some point maybe some people better than me will take a stab at it, and we’ll get somewhere.

From Protein Structures to Larger Biological Systems

Brandon Anderson [00:19:07]: What do you think, Sal?

Sal Candido [00:19:09]: It’s interesting because these models are quite useful, but they’re not exactly the problem that most people want to solve. They solve a very specific purpose, and you can also use them to do other things. For example, you can use them to design new proteins, which isn’t necessarily what you would start with.

I think of the models we’re building now as someone who’s trying to understand how a bicycle works, but is modeling a spoke on it. Those models can get better and better over time, but what you really need to do is move from models of spokes to wheels to whole bicycles, because that’s what people want to understand. To continue the analogy, it seems like people want to use a model of the bicycle to design a part for a pickup truck.

As we put these models into the particular biological context in which they’re operating, we’ll be able to learn more about these interactions on a broader scale. That’s where I think things are going. But I agree that we should keep working on folding models, because they’re going to keep getting better.

Protein Design vs. Scientific Understanding

Brandon Anderson [00:20:43]: With regard to design, there’s a famous Feynman quote: “That which I cannot create, I do not understand.” Now we’re in a world where it’s really easy to create things without understanding them at all. How important is it to have models that help humans understand things, versus magical black boxes that can effectively one-shot a picomolar binder or something like that?

Sal Candido [00:21:18]: We were designing things with magical black boxes long before AI came around. In some sense that’s still useful. But understanding is really important, and it’s one of the big things I think about with AI.

There’s so much for these models to learn. As intelligence gets cheaper and more available, you can deploy it to learn more and more about what’s going on in the world. But how do you pull that knowledge out of the machine so that I can understand it? Maybe that’s just my esoteric curiosity. I think there are so many things to learn.

Brandon Anderson [00:22:05]: A lot of scientists really want to understand things, and the endpoints may not be as important to them. But we’re here to solve translational medicine as a problem, right?

Sal Candido [00:22:18]: I think the more you dig into things, the more you find the right way to keep pushing them forward. One thing that’s really salient to me is that these models have a lot more information in them than we know.

We’ve worked a lot on interpretability for our models, for example, and you find a lot of information there. People know that protein language models learn some notion of structure within their representations, but we find information about functions and motions as well. There’s a lot still to be unlocked, even from the models we have now. That’s important for us to understand as we raise their capabilities.

At the end of the day, you expect a world model to emerge from compressing all this information into a model. How does it do its job? How does it design a protein? It has compressed information from evolution into that model. In addition to being able to produce something useful to us, there’s certainly something to learn just by looking at what’s inside.

AlphaFold Confidence, Calibration, and Interpretability

Brandon Anderson [00:23:48]: What do you think, Pushmeet? Design versus understanding?

Pushmeet Kohli [00:23:53]: I have a different take in the sense that some level of understanding is necessary. Let me explain what level I mean.

AlphaFold isn’t perfect. AlphaFold 2 had a GDT score of about 90 on that set at the time. But even if it had a GDT score of 95, if its pLDDT score were completely uncalibrated, who would trust it? Imagine that it magically gave good answers but told you it was very confident, and then you worked on it for the next year only to find out it was completely wrong. Calibration of the uncertainty measure was extremely important.

In that sense, we do understand AlphaFold 2, and we made a lot of progress in understanding how it behaves. That’s different from understanding how it worked internally to find the solution. There I agree with Sal that interpretability asks: Why did it work? Why did it give this answer?

At the highest level, AlphaFold 2 was interpretable in terms of its behavior and ability to generalize, and we didn’t discover everything about that before launching. When we launched AlphaFold 2 and made the weights available, people found that it was a great disordered-protein predictor. It could figure out which elements of the protein are disordered. That shows it generalizes.

But interpretability asks another question: Interpretable by whom? If you’re saying interpretable by a human rational system, with the cognitive and computational limitations of the human brain, then no, AlphaFold 2 is not interpretable. But if you’re asking whether AlphaFold 2 is interpretable to a much larger, more sophisticated model in terms of how it works, maybe it is. We just don’t get it.

As users of AlphaFold 2, we do need to understand what it can and cannot do. Understanding its behavioral characteristics, strengths, and limitations is extremely important. We can’t just use these models without that characterization, because otherwise, rather than being helpful, they can harm us.

Brandon Anderson [00:26:53]: So your take is that interpretability, strictly speaking, isn’t necessary, but ensuring that a model is trustworthy so humans can make actionable decisions is what people should focus on.

Pushmeet Kohli [00:27:04]: Exactly. Interpretability is also in the eye of the beholder. Who is interpreting it? If it’s a human scientist trying to interpret how the model makes a prediction, that’s a different question from giving another, much larger LLM access to the activation layers and saying, “Can you predict what AlphaFold will do?” Maybe those frontier models of the future will be able to predict that and come up with a theory of how AlphaFold 2 was interpreting and producing its results.

When Will AI Transform Drug Discovery and Clinical Medicine?

Brandon Anderson [00:27:43]: We have a two-minute warning, so one last question for both of you. There’s a real chance that AI will make dramatic improvements in human health in the immediate future. I like quantitative predictions. Best guess: How long until we start seeing AI results in the clinic? Pushmeet, do you want to start?

Pushmeet Kohli [00:28:08]: I think “AI results in the clinic” is an ill-posed question. AI is already being used today in every part of the drug discovery process. From that perspective, it’s already there.

But if you’re asking when we’ll see a 10x or 100x acceleration in timelines, then the next question is: What are we accelerating? Is it lead optimization? Target discovery? Preclinical work or toxicology?

Over the next few years, and this is why the effort announced today is extremely important, we need to tackle some of the hard challenges of biology. Only then will we get the true unlocks of acceleration that dramatically transform drug discovery. AI will continue to be used in things that go into the clinic all the time, but larger acceleration will only be unlocked with a better understanding of the biological models this effort is trying to create.

Brandon Anderson [00:29:43]: All right, thanks. Sal, a quick answer, if you can. We’re almost out of time.

Sal Candido [00:29:47]: I’m tempted to literally put on my biohacker hat to answer this question. It’s important to think about what it means to push the field forward. What we really want to see is outcomes being affected. When will a drug be made entirely by AI? I don’t know. That’s hard to predict. But I do think we’re going to see rapid progress very quickly, because all these tools are already being used.

Going back to your point about basic research and basic understanding, one thing I learned a long time ago in my career, back at Google, is that sometimes it’s easier to approach a problem by asking what it would take to make a 10x improvement rather than a 10% improvement. That’s not because the 10x path is necessarily easier. It’s because it allows you to take a broader view and see solutions you haven’t been approaching. You go back to first principles and ask, “How are we going to do this? How are we going to really push this?”

Both approaches, the 10% and the 10x, are valuable, and we should do both. I’m very happy that at Biohub we have a beautiful and lofty mission statement: to cure all disease. If you want to do that, you really need to figure out what the 10x approach is. It’s a great opportunity to be able to go do that.

Brandon Anderson [00:31:46]: Awesome. Thank you both. Thank you for being in the literal hot seat. Very interesting.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit https://www.latent.space/subscribe

It’s hard to believe that Periodic was only launched last September:

One year later, it is considered one of the pre-eminent AI scientist labs, with dizzying talent density and astonishing progress in the autonomous lab buildout:

Most people are familiar with the standard credentials of Liam and Dogus, but we found an incredible “talent slope” while learning more about Periodic, where each successive employee seems more impressive than the last:

From building AI systems that reason over noisy physical experiments to creating laboratories where every instrument can become intelligent, Periodic Labs is betting that the next frontier of AI won’t come from simply training on more internet data, it will come from letting models experiment with the real world. In this episode, Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk join swyx and Brandon to explain why scientific discovery is fundamentally different from math and coding, and what it takes to build AI scientists that can actually discover new materials.

We go deep on Periodic’s vision for “synthesis superintelligence”: reinforcement learning grounded in physical experiments, AI-powered materials characterization, simulations and density functional theory, high-throughput labs, and systems that learn from the entire process of doing science rather than only its published results. Liam and Dogus also explain why frontier models will still need experiments, why failed experiments may be some of the most valuable training data, what it means to give every piece of lab equipment “140 IQ,” and how autonomous experimentation could compress decades of scientific trial-and-error into months.

We discuss:

* Why intelligence alone isn’t enough for scientific discovery

* How reinforcement learning changes when the environment is the physical world

* Why science requires reasoning under uncertainty, noise, and missing information

* Prediction, synthesis, and characterization in the materials discovery loop

* Why physics and materials science are still far from “solved”

* The “matter compiler” and Periodic’s goal of synthesis superintelligence

* Phase transitions, X-ray diffraction, and AI-powered materials characterization

* DFT, simulations, and why experiments remain the ultimate ground truth

* Room-temperature superconductors, new magnets, batteries, and more efficient compute

* Why quantum computing may not automatically solve materials discovery

* What it means for every piece of lab equipment to have “140 IQ”

* Why data quality and negative results matter more than simply throwing more compute at science

* Training models on the process of doing science rather than the final answer

* Why even future frontier models will still need to physically experiment

* Scaling autonomous labs across AI, chemistry, physics, and custom hardware

* How automated experimentation could massively increase the “surface area for luck” in discovering new materials

Liam Fedus

* LinkedIn: https://www.linkedin.com/in/liam-fedus-26547811/

* X: https://x.com/LiamFedus

Ekin Dogus Cubuk

* LinkedIn: https://www.linkedin.com/in/ekin-dogus-cubuk-9148b8114/

* X: https://x.com/ekindogus

Timestamps

00:00:00 Introduction

00:02:49 AI and Reinforcement Learning in the Physical World

00:09:17 The End-to-End Materials Discovery Loop

00:13:28 Why Physics Isn’t “Solved”

00:19:04 The Matter Compiler and AI Characterization

00:30:22 DFT, Simulation, and Experimental Ground Truth

00:42:27 Dream Materials, Compute, and Quantum Computing

00:45:57 Giving Every Lab Instrument “140 IQ”

00:50:02 Automating the Lab

00:53:16 Data, Models, and Negative Results

00:58:13 Training AI on the Process of Science

01:00:20 Why Frontier AI Still Needs Experiments

01:06:06 Scaling Autonomous Labs

01:10:45 Building the Team and Deploying to Industry

01:17:01 From AI Copilots to Scientific Outcomes

01:20:57 Automating the Search for Superconductors

Transcript

Introduction: Intelligence Must Meet Reality

Swyx [00:00:00]: Okay. We’re here at Periodic. We’re here today with Liam and Doğuş from Periodic. Welcome, and thanks for having us at yours.

Liam Fedus [00:00:11]: Yeah, great to be here, and yeah, thanks for coming in.

Swyx [00:00:13]: I want to start off with one of these quotes that I love from your website. It says, “Intelligence is necessary but not sufficient. New knowledge is created when ideas are found to be consistent with reality.” It seems so straightforward, but why is it non-obvious?

Liam Fedus [00:00:29]: I think that’s sort of the thesis and premise behind Periodic that Doğuş and I thought when we were building this, which is you can’t just think your way to a solution. The universe is so complicated that in order to actually push the frontier of knowledge and to make progress, you need to create these conjectures and then actually see whether or not it holds. No amount of rereading that textbook or paper, thinking, or disappearing into a room is going to allow you to think through all possible experimental outcomes, expected and unexpected, and you really need this iterative process. And that’s our core belief for putting this together, and that’s why from the very beginning, we’re like, “We need to have the AI systems, the simulations of the physical world, but then also to build up the physical high-throughput experiment.” And we think a different type of intelligence emerges from that.

Swyx [00:01:21]: I think it’s also notable, like with OpenAI and Google, you don’t have those resources in those big labs. You had to come out and do your own thing, because you’re kind of inventing your own playbook as you go along.

Building an Interdisciplinary Lab for AI Science

Ekin Doğuş Çubuk [00:01:34]: Yeah, and a team like this has never existed, I think. We try to bring together solid-state chemists, solid-state physicists, experimentalists, theorists, hardware engineers, LLM experts, computer scientists, and some of these technologies are very recent. The high-throughput experimentation, the robotic arms have only been tried in the last three years or so. The force field expertise, some of these force fields are also very recent. But yeah, we felt like having a lab is very important. Having this focus and bringing all these people together to work together is very important. And I guess the closest examples from history are places like Bell Labs, where incredible theorists and experimentalists and chemists worked together and achieved incredible things. So we are trying to do the same here, yeah.

Physical-World Reinforcement Learning and Uncertainty

Brandon [00:02:17]: I want to talk about force fields in a bit, but

Swyx [00:02:19]: Yes

Brandon [00:02:19]: Before we get to that. Yeah what are they? But before we get to that, For context, some of our audience are AI engineers, and some are scientists. For the engineers: what’s the difference between working at a big lab like OpenAI and what you’re doing here? I think you’ve already hinted at this but maybe explicitly, how do you have to reframe your thought process?

Liam Fedus [00:02:49]: Well, I think one interesting thing is now our reinforcement learning environments literally derive from the environment, from our physical labs. Our data comes from our physical labs, and this is sort of our ultimate truth. It’s not enough just to do optimization against, some answers that were known in some papers or textbooks because we’re going beyond that, and I think that’s one of the biggest differences. But along with that, there’s issues of say, decision-making under uncertainty. So when you’re doing optimization against math, there’s a high precision to it. You’re not really dealing with kind of like variance or uncertainties or aberrant measurements, whereas that’s very key to our process. For example, when we are doing a materials discovery loop, things don’t come out of the furnace labeled, right? Even the labeling process can be stochastic and noisy. And sometimes you’ll have aberrations between different machines. Maybe we’ll have insufficient telemetry. So we thought we ran it at this temperature, but really the temperature was, some delta away. And an intelligence that can take in this noisy data and make intelligent decisions like a scientist would, is a sort of a different set of reasoning strategies. So I think there’s a huge amount of commonalities, so the standard mid-training, reinforcement learning, the construction of tool-using agents, variance reduction, making sure infra doesn’t have mismatches between training and inference. But we have to go beyond that and really think about how do you do accurate, work when there’s a high amount of uncertainty? How do you make really incredibly efficient use of a limited set of data? So we’re scaling things up significantly, but still, it’s very different than a digital environment where you can arbitrarily add more environments or more rollouts. We don’t have that same capacity. So sample efficiency is another key thing.

Why Experiments Are Noisy and Incomplete

Brandon [00:04:48]: This seems a little bit abstract to me, and it is a theme which has been on the science pods for a bit, but I think it might be helpful when you talk about experimental uncertainty, to explain what would a specific process that you were doing experimentally look like, and how would a human approach this problem, before we get into the AI side? Do you have a specific example of a type of material you make and what sort of uncertainty in the measurements you would experience in the process?

Ekin Doğuş Çubuk [00:05:17]: I think there are many dimensions to this. So one of them is I think every scientific experiment has to do some dimensionality reduction, and this is very different than coding or math. So, when you’re doing math or coding, all the context could be available to the human or the LLM. It’s a bunch of axioms, some corollaries people have derived from them. It’s a bunch of functions and API calls. So everything you need to reason through is available to you, and then you just have to be very smart and figure it out. In physics, as you know, we start with more atoms than we could ever store on a computer. So clearly, we’ll have to go from the original number of dimensions and bits that represent a system to the number of bits we can fit in the computer. And this was, I guess, the original premise of thermodynamics. They were originally really confused about how steam engines worked, but then turns out you can find five or six thermodynamic variables that explain what’s going on pretty well, which is incredible. That’s the beauty of physics.

Brandon [00:06:13]: You have a room full of atoms, and there are 10²³ or 10²⁷

Ekin Doğuş Çubuk [00:06:19]: Exactly

Brandon [00:06:19]: Or something atoms in this room, and yet you need temperature and pressure and simple variables, and that tells you everything. No, it tells you mostly what you need

Ekin Doğuş Çubuk [00:06:28]: A lot

Brandon [00:06:29]: To know for most cases, right?

Ekin Doğuş Çubuk [00:06:30]: Yeah.

Brandon [00:06:30]: So let’s say you make a material, and you could do something like look at an X-ray structure, and this thing is a very lossy, projection, right? It’s not actually a structure, or it doesn’t tell you the true structure. It tells you some piece of principal components of that. So, how would a human take these simple principal components and then, kind of reason about them? And then how do you extend that to an AI?

Ekin Doğuş Çubuk [00:06:58]: So on the theoretical physics side, right, we’ve decided that energy, the average value of energy and the fluctuations of energy is very important. So a lot of thermodynamics has been derived from this. And then we realized reaction barriers are also very important. The kinetics is very important. So a human can try to reason through this extremely complex system by coming up with some, reduced dimensionality descriptors. People also really like atomistic picture. You think about what the local neighborhoods look like, whether they’re from octahedra, tetrahedra, and that especially chemists really benefit from that. Okay, so that’s on the theoretical side. And on the experimental side, you would try to collect as much data as possible. So you’d have a lab notebook, and you’d write everything you see, et cetera, but with the understanding, of course, that you’ll miss a lot. So like as Liam mentioned, there are all these issues that come up. For example, your furnace will degrade over time because as you’re using the furnace, some of the things you’re baking will evaporate and cover the heating element. So every time you use the furnace, it gets worse and worse. Another issue is you have a furnace, but the temperature isn’t perfectly uniform, so where you put in the furnace will affect the result. We don’t have this issue yet, but we will maybe someday. The optical instruments, suffer a lot from vibrations. So every time somebody walks. I remember when I was doing my PhD, one lab had this big issue because sometimes their results would be very different than other times, and they brought it down to at a certain time in the night upstairs, somebody would be walking and that vibration affected the laser setup. So yeah, science is very hard, but that’s what makes it special, right? The other thing Liam and I have been talking about is a lot of the current improvements focus on math and coding and theoretical computer science because it’s easier for LLMs. But in real life, most things that require intelligence are actually more like science. There’s a lot of uncertainty, there’s a lot of noise, there’s a lot of missing context, but you have to be the intelligent being and figure out what to do next.

Liam Fedus [00:08:53]: I think a really interesting point to build on that too is the optimal reasoning strategies to do math, to do theoretical computer science, to do these types of things may not be the optimal reasoning strategies to do science.

Ekin Doğuş Çubuk [00:09:04]: Yeah.

The Materials Discovery Loop: Synthesis and Characterization

Brandon [00:09:05]: How do you define an RL reward function when your output is a noisy crystal structure or some number of I don’t know collective variables or something? Like.

Liam Fedus [00:09:17]: Maybe, I’ll give, one part of the loop, and I’ll just kind of dig into that. So, as part of the materials discovery loop, we have to first figure out, what to make. So that’s the stability of the atoms together, as well as do we expect it to have the properties we want. So we don’t just want, a novel configuration of the atoms. We want them, when put together, to have the property of interest. Next is how do you synthesize it? How do you actually make the thing? That’s also non-trivial. So even if you know something will hold together, the process of actually, what are the processing conditions to actually bring that forth is highly non-trivial. But again the thing. Once you’ve done those two steps and you’ve made something, it doesn’t come off labeled, so you actually have to characterize it and figure out what you made. So in that case rather than thinking about the RL environment of we’re just going to kick off an experiment, wait a couple days, and try to do a update on did you find a room-temperature superconductor or not, that’s completely infeasible.

Brandon [00:10:15]: Yeah. I was going to ask that. Yeah you can’t leave your GPU sitting for a week or something.

Liam Fedus [00:10:19]: It’s. Yeah, it’s completely infeasible

Brandon [00:10:20]: Yeah

Liam Fedus [00:10:20]: Because you don’t have enough agents to reduce the variance sufficiently. The rollout time is going between the different instruments. You’re waiting. There’s just some physics limits of the experiments, for the actual synthesis or furnace time. So it’s just too slow, too noisy. So the way we think about the AI program is basically constructing agents around this set of data that we’ve produced. And so in the case of characterization, you’re looking for RL environments that are rewarding the identification of the phases actually present. Are you able to given the raw experimental data, fit that effectively? So the way we do this is we’ll shoot X-rays, at the material. X-rays have a frequency or a wavelength roughly measurable or commensurate to the spacing between atoms, so you get nice diffraction patterns that kind of give a fingerprint of the crystal structure. And doing this, identification of what actually is present is highly non-trivial. And so some of our early AI work has been on basically producing systems to do this. And this allows us to move much more quickly too, because if you’re running so many experiments, you really quickly become bottlenecked on your ability to understand what was produced. But again, now the reinforcement learning environment is fairly straightforward, which is you’re going to reward the identification of the phases present. You’re going to penalize spurious phases or things that don’t have any real chemical plausibility. So that’s like a little succinct thing where you can kind of do this for the different pieces. And when you stitch this together, that’s the end-to-end discovery loop.

Training on Novel Experimental Data, Not Memorized Answers

Brandon [00:12:02]: Okay. You

Liam Fedus [00:12:02]: Maybe another

Brandon [00:12:03]: Yeah. Oh, sorry

Liam Fedus [00:12:03]: Another piece would be as we build up, a large basket of experimental data, you can think of another thing of basically timestamping the state of the world. So you could say, “At this date, this was our experimental evidence up until that point.” And you can construct reinforcement learning environments where you say, “Okay, given that body of experimental evidence and given, the choices, what was the next choice that the scientist made, or what was an outcome of that experiment?” And. This is really interesting too, because this allows us to do types of programs that would be harder externally. Because if the pre-trained model has already memorized some of this data and then you try to do reinforcement learning on that, if it already knows the answer and you construct reinforcement learning tasks that depend on that answer, it can fake work. So, and then, because it’s like, oh, well, it already knows the answer, so it doesn’t actually have to do the hard physics reasoning. It doesn’t have to actually do the calculation simulations. And it gets to the right answer, and then you reinforce that policy and you up-weight those reasoning strategies. Those reasoning strategies will not generalize to novel systems, so it’s not useful to us. But as we build up baskets of experimental data and stitching things together, having the lineage through the whole process, that gives us the ability to construct new types of environments that are just not plausible elsewhere.

Why Known Physics Does Not Mean Perfect Simulation

Swyx [00:13:27]: I want to ask my question

Brandon [00:13:28]: Seriously

Swyx [00:13:28]: But I’m worried that I’m taking it off the rails. I’m just going to go for it because I, again, I come from the non-scientist, but engineering sort of. I’m familiar with the ML side, but not the not the physics side. Okay, there’s a few versions of this, but I think the basic question is, don’t we have most of the laws of physics worked out? And how come we don’t have perfect simulators already?

Brandon [00:13:50]: Oh, that’s a great question.

Swyx [00:13:52]: Like, this is very basic, like. And he was like, “Ha, that’s cute.” But like

Brandon [00:13:56]: No, it’s a great question

Swyx [00:13:57]: Dude like I have all these physics books. What do you mean, what do you mean we’re not done?

Ekin Doğuş Çubuk [00:14:03]: Yeah, so there are a couple of things going on here, right? So

Swyx [00:14:07]: Yeah

Ekin Doğuş Çubuk [00:14:07]: I think first of all, we’re definitely not done with any of the laws of physics. Even the

Swyx [00:14:12]: At quantum scale, okay, but. And maybe very large scale, but at human scale, at material scale, we’re, we. What’s left?

Ekin Doğuş Çubuk [00:14:22]: We’re not. So it’s a it’s a really good question. And it’s very interesting why we’re not done yet, right? So there’s a very, common story people talk about. When Dirac was figuring out quantum mechanics, like late 1920s, he wrote a textbook about quantum mechanics, and he kind of framed it as it’s all done. And then there’s a phrase people like to quote. I’m not sure actually how accurate it is, but apparently Dirac said, “The rest is chemistry.” The idea being he could solve the hydrogen atom

Swyx [00:14:50]: Just compose everything.

Ekin Doğuş Çubuk [00:14:51]: Yeah. He could maybe solve a chain of 1D hydrogen atoms, but then he cannot currently solve, say, nitrogen interacting with oxygen, but that’s okay, that’s just chemistry.

Swyx [00:15:00]: Yeah.

Brandon [00:15:00]: Physics physicists really love single atom rep. Really simple systems.

Ekin Doğuş Çubuk [00:15:05]: Simple systems.

Brandon [00:15:05]: And that’s all you can solve.

Ekin Doğuş Çubuk [00:15:07]: Spherical cow.

Brandon [00:15:07]: Yeah, spherical cow.

Ekin Doğuş Çubuk [00:15:08]: But it turns out, I think what we learned in the last hundred years is, first of all, that’s not true. It’s not like a simple extension of what it was to do just hydrogen. And it’s not just. The rest is not just chemistry. There’s actually a lot of physics there. I guess one of the things we still haven’t figured out is high-temperature superconductivity. But there are a lot of other

Swyx [00:15:25]: And high for you is 175, 200?

Ekin Doğuş Çubuk [00:15:28]: Oh, when people say high-temperature superconductivity, what they mean there is unconventional superconductivity. So there’s the conventional superconductivity that’s just mainly driven by electron-phonon coupling. And in that case, you can kind of see the isotope effect. If you take the same system but just different weight for one of the elements, you can see the superconducting temperature drops at the rate you expect. So that one is conventional. But then higher-temperature superconductors like cuprates, like the ones that are above 77 kelvin, like 93 kelvin, turns out don’t obey that physics. But we don’t know what physics they obey. They’re just incredible superconductors. But it’s not just that. So that definitely is one of these very, popular topics we don’t know the theory for yet. But there’s so many other things we don’t understand. For example, even strong correlation in simple quantum mechanical systems, we cannot simulate yet. Density functional theory is a very powerful tool. It’s incredibly accurate on some things, but as long as. As soon as there’s some strong electron correlation, it actually fails to capture some of the effects. There’s so much to figure out. That’s why I’m really excited for AI to be applied to this field, because theoretically, computationally, and experimentally, there’s so much to figure out. And every improvement we make here should improve human life, because, the better we can understand materials, solid-state physics, the better devices we can make for them.

Brandon [00:16:41]: There’s this famous quote by Phil Anderson, which is, “More is different,” which is basically you may understand all the basic laws of something, but when you add many things, they behave qualitatively distinct, differently from how the simple physics should tell you. So understanding how lots of things behave together, it seems like it should be simple. Beginning laws are simple, but the collective behavior is so complicated that it’s really hard to model.

Ekin Doğuş Çubuk [00:17:07]: It’s emergent and universal, which is crazy. And we’re also seeing this with deep learning models, right? We see power laws everywhere in deep learning. We don’t understand, but it is very reminiscent of physics where there is, emergence and universality.

Brandon [00:17:20]: There’s been many theory papers in the physics world about emergence in neural networks.

Ekin Doğuş Çubuk [00:17:24]: Yeah, absolutely, yes.

Brandon [00:17:25]: Yeah.

Synthesis Prediction and the Matter Compiler

Swyx [00:17:26]: And then my other follow-up question was also just on the process of let’s call it the yeah, process and the end result. Let’s call it like, you want to have these properties that you’re targeting for materials, and then you have to figure out the process to get there. Is it worth separating these things so that you can have. You just train a model that perfectly reverse engineers any process whatsoever given, whatever theoretical end state you want? Is that, meaningful?

Ekin Doğuş Çubuk [00:17:54]: Maybe going back to the conversation we had about how since there are more than Avogadro’s number of atoms, we’ll never know the exact context of what happened. I think that might be a reason why we’ll never be able to reverse engineer everything perfectly. But we’re just trying to reverse engineer sufficiently to be able to improve the performance of the materials, basically.

Swyx [00:18:13]: Yeah.

Brandon [00:18:13]: What you’re asking is there a forward model where you put in an input and you can predict the output, and you want to under

Swyx [00:18:19]: Yeah.

Brandon [00:18:20]: You want to create a like a surrogate model which can understand

Liam Fedus [00:18:22]: The input is the final state of that, this configuration of atoms, and then the prediction is

Swyx [00:18:27]: Because then you can, you can split up the work almost. You can have one team do the process side, and then the other team do the property side, and then

Liam Fedus [00:18:33]: Yep

Swyx [00:18:33]: You know, just race them.

Liam Fedus [00:18:36]: Yeah, I think that there’s, a lot of progress we can make in terms of these. Splitting up these agentic workflows to these different areas, I think there’s a lot of empirical work we can make, on synthesis prediction as well.

Swyx [00:18:49]: Yeah.

Liam Fedus [00:18:49]: Whereas taking in that configuration of atoms, what tools can you use? How, given this experimental evidence, given prior literature, prior papers, how do you actually get to these things? And you ultimately are testing, are you able to actually replicate it or not?

Swyx [00:19:04]: The reason I say this maybe also is just thinking about it in terms of a company. You know, in the way that TSMC is kind of like a fab for semis, but they don’t design the semis, you could be the TSMC of materials where people come to you with like, “Well, here’s what I want,” and then you can figure out the process and

Liam Fedus [00:19:21]: Yeah, it’s like a matter compiler.

Swyx [00:19:23]: Yeah.

Liam Fedus [00:19:24]: So I guess given these requirements

Swyx [00:19:26]: Very Star Trek.

Liam Fedus [00:19:27]: Yeah. But it’s like, given these requirements, is this actually a valid set of configurations? Can this, can this exist?

Swyx [00:19:33]: Is it a good objective function? I don’t know. It’s too big.

Liam Fedus [00:19:37]: Yeah. We kind of talk about one of our missions as synthesis superintelligence, so I think it’s in line with this.

Phase Transitions, X-ray Diffraction, and RL Rewards

Swyx [00:19:44]: Yeah.

Brandon [00:19:44]: I want to double-click on something that you said a little while back, which was you talked about phases. And I think this is, going to Doğuş’s comment about, what is emergence. The concept of a phase transition I think might be foreign to a lot of the audience. So can you explain what is a phase transition, and then why is this a helpful signal for something like an RL environment?

Ekin Doğuş Çubuk [00:20:08]: Phase transitions are fascinating. I think the maybe the one that people can most easily relate to in their life is, ice melting probably, or water boiling. You increase the temperature of ice, and it still looks like ice and it behaves like ice, but there’s a certain temperature at which it just stops raising its temperature and then turns into liquid. So it suddenly goes from this solid phase to liquid phase. Another one that’s very relevant to us, of course, is the Ising model. And I think computer science and mathematicians also study this maybe under different names, but basically, you can have spins up and down, and they have different interaction terms if they’re both up versus one is up, one is down. And at high enough temperature, they’re usually just randomly up or down. Entropy wins. You lower the temperature, it’s still looking the same. But then at some point, suddenly they perfectly start aligning. Yeah, the phase transition. Phase transitions are really nice because it makes physicists’ life easier for studying certain things. They tend to have certain, spatial correlations that really help us, as well. Okay, so in our lab, phase transitions come about usually because we mix the precursors that are different, crystals basically, but then we raise the temperature or do something else to encourage them to react, and then atoms start reacting and they form into this new crystal. And the this shows up in the XRD pattern because usually if the geometry of the atoms change drastically, the XRD pattern changes a lot. But what really happens in a typical practical materials discovery campaign is when you first try something, it doesn’t work, and the outcome isn’t just a clear yes or no, but it’s usually a very mixed phase. It will usually have some of the precursors. It might have some amorphous phase, and then it will have a bunch of phases that you maybe didn’t try to make. So that becomes a challenge, and that’s why we find it very difficult to bring together simulations and AI to do the characterization. So the simulations can always say, “Oh, this phase doesn’t look like what we predicted, but this is a slight variation of it. Let me now do force field calculation on it to see if it’s still stable.” Or the AI can say, “Based on the previous experiment and this one, this is probably not the phase we want to make, so let’s change the conditions.”

Liam Fedus [00:22:18]: Yeah, and in some cases

Swyx [00:22:19]: For listeners, XRD is X-ray diffraction, which you already described.

Liam Fedus [00:22:22]: Yeah. In some cases, there’s a phase that just hasn’t been recorded before. It doesn’t exist in any papers or database. And so the AI has to actually go and sort of use these tools to say, “Well, what configuration atoms would actually explain these phases?” But this becomes incredibly important for directing the scientific process because, let’s say we have a phase target in mind. We really need to sort of hill climb that to get better measurements on that. So if it’s, one percent and we’re like, actually, we want to increase this phase purity, we want to have a very good measurement on this experimental data to say, what is present and what’s not.

Brandon [00:22:57]: So one of the advantages is you have just a distinct variable that you can just observe. Going back to the statement about how do you deal with uncertainty, one of your, it seems like part of your answer is you make your observable signature somewhat unambiguous. There’s no interpretation there. Is that right? Is that the logic behind this? Or is this literally you just needed to find a new phase, and so this is the thing you care about?

Liam Fedus [00:23:20]: I think there’s still ambiguity in a lot of these cases. A very dumb thing is replicates.

Brandon [00:23:25]: Yeah.

Liam Fedus [00:23:26]: We run replicates, of course.

Brandon [00:23:27]: Yeah. Yeah.

Liam Fedus [00:23:28]: But no, I think there — You know, it’s a challenging task because it’s not just a fully deterministic, and there are well, there are ambiguities, and other things. Two different, phases can actually be consistent with the same pattern.

Brandon [00:23:42]: Yeah.

Liam Fedus [00:23:42]: And so I think that’s why it’s really important for the system to use some chemical intuition to say “Okay, well, this would be highly unlikely given the synthesis conditions, given this, the prior knowledge.” and so that can be used to help disambiguate.

Chemical Priors, Multimodal Measurement, and Ambiguity

Swyx [00:23:58]: So you’re, so you’re somewhat injecting priors there, based on what you expect.

Ekin Doğuş Çubuk [00:24:03]: Yeah. I think thermodynamics is the biggest prior, right?

Swyx [00:24:06]: It is.

Ekin Doğuş Çubuk [00:24:07]: And then physics is a big prior. Another one that we really benefit from here is multimodality or materials characterization. So we can do XRD, and yes, some phases might look similar on the XRD pattern, but then we can also measure their electrical properties, magnetic properties. We can measure their morphology, like some kind of electron microscopy. And then some of the phases that look similar on XRD will look different in some of these dimensions. And that kind of multimodality really helps. And again, AI is really helpful here because humans have limited context and computation power as well. So if you give humans 10 different modalities at the same time and say, “Analyze these all consistently,” it’s a bit difficult. But it’s actually not super intelligent for an AI to do it. It’s great, yeah.

Liam Fedus [00:24:46]: Yeah, it’s like the super intelligence is it can just do so much calculations. It can look at everything. And this is, kind of gets to that, phrase about better decision-making under uncertainty, where now it’s stitching together, the signal across many different instruments longitudinally, across different experiments, across the replicates, and then kind of getting to this underlying better model

Swyx [00:25:08]: Yeah

Liam Fedus [00:25:08]: Rather than just over-indexing on one measurement from one instrument.

Information Limits, Telemetry, and Maxwell’s Demon

Swyx [00:25:12]: And let me just, like. So you know, you started, I think a while ago also saying that there’s just more data than you can fit in any reasonable computer or process or anything. So at some point, you have to throw out data even though you have all these replicates, even though you have presumably ten different ways to measure the same thing. It’s. ‘Cause what if one of your temperature things is wrong? So when do you throw that out? How do you decide what to do? As a data-hungry ML guy, I just want everything, right? And then I just, you know

Ekin Doğuş Çubuk [00:25:37]: So it’s, yeah.

Swyx [00:25:37]: Waddle, learn everything. I don’t know what I don’t know, and it’s just throw it, throw it, the machine at it.

Ekin Doğuş Çubuk [00:25:42]: Yeah. So we don’t throw any data. I think maybe what I meant was potentially. If God was observing the experiment

Swyx [00:25:50]: Yeah.

Ekin Doğuş Çubuk [00:25:50]: There is more data in the experiment than humans can fit in a computer. Unfortunately, us mortals cannot observe all the atoms. So even though there’s more data than we could store, we actually can’t access that data. For example, I cannot trace all the atoms’ positions. So unfortunately, we’re still in a regime where any data we collect is very precious, and we don’t delete any of it. Yeah.

Brandon [00:26:11]: Maybe a way of thinking of it to connect to the AI side is think about probing, like linear probing of a of a foundation model, right? This model has all this data, so God has this entire picture of what the material is doing, and then the experiment is this little, tiny linear probe which has eight bits or something of output. And so you’re taking this humongous thing and coarse-graining it down to a little bit of information, and then it’s now, either traditionally a human’s job or now your agent’s job to reconstruct from those eight bits of information

Liam Fedus [00:26:43]: It’s an analog, yeah.

Brandon [00:26:44]: What is actually happening inside.

Liam Fedus [00:26:45]: It’s an analog where you can’t have full visibility over every parameter weight, over every activation for that input. Yeah, you’re getting some sort of subset on it or some coarse-grained features across the whole thing.

Ekin Doğuş Çubuk [00:26:58]: Yeah. But so you’ll probably find that. You probably, do find, Maxwell’s demon argument very interesting because that kind of

Swyx [00:27:05]: Can you, can you res-recite it? Because I don’t. He’s mentioned it before. I keep forgetting it.

Ekin Doğuş Çubuk [00:27:10]: So, in the 1800s, people realized that entropy of a system has to increase or a for a closed system

Swyx [00:27:18]: Yeah.

Ekin Doğuş Çubuk [00:27:18]: Or stay the same, but it can’t go down.

Swyx [00:27:20]: Yeah.

Ekin Doğuş Çubuk [00:27:21]: It’s the second law of thermodynamics. And there was a thought experiment that was asking, what would happen if there was a very tiny, all-knowing demon that could kind of look at atoms, and then whenever an atom had low energy, kind of open a door and let it into this other room and keep doing this to reduce entropy of the system, which would break second law of thermodynamics? And for a long time, I think there wasn’t a very satisfying answer for why that was the case. But then there was a more recent paper in the 1900s, more recently to us, Landauer came up with the argument that you’d have to spend energy to delete information, and Maxwell’s demon would basically have access to so much information by doing this thing he’s doing that he’d have to delete some. And for. When he’s deleting information, he’d have to spend energy, which would then make sure entropy increases. So even though we’re not at the level where we can be Maxwell’s demon, I think your, question is very visionary towards a future where we store so much data we have to delete some.

Swyx [00:28:17]: Yeah, no, I wasn’t, I wasn’t obviously going to that, to that extent. It does remind me of air conditioning. That’s kind of how air conditioning works. But also, I think it’s just a question of why don’t you just buy every sensor in the world and just over. Ridiculously over-instrument, over-replicate everything?

Liam Fedus [00:28:33]: Oh, we do that.

Swyx [00:28:34]: Right? Yeah, okay, yeah, okay. There we go.

Liam Fedus [00:28:35]: Yes, yeah.

Swyx [00:28:35]: Okay, it works out. Right. All right.

Liam Fedus [00:28:37]: Yeah, I think, I think it’s a matter of

Swyx [00:28:39]: Confirmed.

Liam Fedus [00:28:39]: Yeah, exactly. Are you getting visibility to individual atoms, which is infeasible? But no, we add a huge amount of telemetry, and that basically reduces the number of hidden variables, if you will.

Swyx [00:28:49]: Yeah. Yeah. And just out of curiosity, you guys do you worry that you’re just in California? Don’t you want to do this in Nepal and Australia?

Ekin Doğuş Çubuk [00:28:59]: Yeah, we could. We have plans to open more labs.

Swyx [00:29:02]: Because obviously gravity and where you face and where you. Where you are matters.

Ekin Doğuş Çubuk [00:29:07]: Also expertise can change, right? Here we have certain kinds of expertise. For example, Stanford Physics Department, Engineering Department has certain kinds of expertise that really helps us because we can hire from there. Our researchers can collaborate. As you said geography can matter, but then different cities can have some differences of expertise within physics, chemistry.

Swyx [00:29:27]: Yeah. Oh, we have Zoom

Ekin Doğuş Çubuk [00:29:28]: Yeah.

Swyx [00:29:28]: For that.

Ekin Doğuş Çubuk [00:29:29]: Yeah. But still

Swyx [00:29:31]: It’s not like your experiments are going to be affected by where you are.

Liam Fedus [00:29:34]: Yeah. Yeah, the future of Periodic is not a single lab.

Swyx [00:29:38]: Yeah. You gotta head up your private jet and go to space and you know.

Liam Fedus [00:29:42]: Right. Yeah, I think it’s necessary to have this sort of network of labs. And like, yeah, I think as Doğuş was just kind of pointing out there’s an optimal place for each of these based on expertise or materials constraint. There’s many different constraints, permitting.

Connecting Simulations With Physical Experiments

Brandon [00:29:56]: Okay, so one of them is. So one thing I-I’ve been curious about is how do you deal with computational tools? Or how do you integrate computational tools into a decision-making workflow, especially when, let’s say, computation and experiment don’t necessarily, meet? It’s also theory in terms of high-level understanding. And yeah, how does that sort of work together?

Ekin Doğuş Çubuk [00:30:22]: Yeah. So, one thing that’s really helpful here is there are certain things that are easier to do in simulation and more accurate in simulation. There are certain things that are easier to do in experiment and more accurate in experiment. In general, of course, experiments are more accurate, but certain measurements are very hard to make with experiment. So you make a bad version of it, so you get inaccurate data. Let me just give you some examples. So one of them is, we use density functional theory to estimate formation enthalpy of materials and

Brandon [00:30:51]: Sorry, quick pause. Can you explain densital density functional theory and formation entropy?

Density Functional Theory and Formation Enthalpy

Ekin Doğuş Çubuk [00:30:55]: Yeah, of course.

Brandon [00:30:56]: Sorry, enthalpy, yeah.

Ekin Doğuş Çubuk [00:30:57]: Yeah. So, density functional theory is probably the most commonly used simulation method for materials. It kind of comes from Hohenberg–Kohn theorem, and Kohn got a Nobel Prize for this. He. When he realized that, sure, quantum mechanics is very expensive because you basically have this exponential Hilbert space for every wave function, so you have to solve this very difficult problem, especially as the system size grows. Kohn and Hohenberg realized that turns out you don’t need to do it at exponential space. All you have to do is think about the charge density, at least for the ground state of quantum mechanical systems. So this it’s actually the theorem is pretty simple. So even I think an undergraduate in physics can understand. But the idea is all the properties of a quantum mechanical system ground state is a functional of the charge density, which is amazing because charge density is just this three-dimensional object, whereas the Hilbert space and wave functions are exponential. So that was a really interesting observation. And then Kohn published another paper, I think with his postdoc, Sham, Kohn–Sham, wave functions, kind of had a practical way to solving some approximation of quantum mechanical properties of materials. And today, we use it very much, both in our lab and externally. And then formation enthalpy is one of the measurements we can make to understand the energy of the material. Basically, we want to ask, what is the energy that this material has? Because if that energy is higher than other materials these atoms can come into, then it’s unlikely that this material will be able to be made. So stable just means like it’s on the convex hull of other materials, and their energies. Was that clear enough, or should we.

Brandon [00:32:42]: Maybe a simple high-level point. So density functional theory is a way of taking something which is exponentially hard, in number of electrons

Ekin Doğuş Çubuk [00:32:51]: Right

Brandon [00:32:52]: Or number, in your system size, and then reducing it approximately with some introduced error to maybe usually an n-cubed approximation. So now you can compute quantities which are, you know. You can now compute things which are just uncomputable, but at some cost, right? If you could do it perfectly, you would — we wouldn’t need a lab. But we can’t. This might be a good callback to our, actually our second episode ever. Heather Kulik had said, she famously said, “There is no AlphaFold for materials, not just computationally, but the ground truth just doesn’t exist.” Like, we don’t really know what material crystal structures look like, and DFT is generally our best route towards.

Ekin Doğuş Çubuk [00:33:34]: Yeah, I think so. Maybe two things I would add is DFT by definition doesn’t have to be an approximation. The Hohenberg–Kohn theorem shows it

Brandon [00:33:42]: It doesn’t

Ekin Doğuş Çubuk [00:33:42]: It can be exact. But I think, as you’re pointing out, the exchange correlation functional we don’t yet have access to. And then for the charge density-based models, we don’t know what the kinetic energy functional is. And then even if we had perfect DFT, I still think we’d need a lab because we. Even if we had perfect DFT, we cannot fit ten to the twenty-three atoms in a computer.

Brandon [00:34:03]: Yeah.

Swyx [00:34:04]: You’re very obsessed with that number.

Brandon [00:34:06]: But. No. Well, okay. Yeah. But with exponentially growing compute, you can argue that in n-cubed scaling size, eventually you could, in principle, calculate things. You just wait till the pause long enough.

Ekin Doğuş Çubuk [00:34:17]: Oh, yeah. We need massive computers, yeah.

Brandon [00:34:19]: But yeah. But as long

Ekin Doğuş Çubuk [00:34:20]: It’ll take a while.

Brandon [00:34:20]: Yeah. But as long as n-cubed is a good scaling, you can, you can.

Swyx [00:34:24]: As the non-scientist, I will make an observation, which I don’t know if you want to throw around or anything like that. So I came from finance. We had the Gaussian copula, which was a way of pricing credit default swaps, by doing correlations and reducing everything into a single kernel. And it’s very. It feels also spiritually similar to the VAE, in terms of like, condensing all these things into a single sort of parameter. I wonder if just there’s the same trick everywhere. Yours sounds, a bit more high-dimensional than mine, which VAE is just like, one sigma thing. But sounds the same?

Liam Fedus [00:35:01]: Yeah, the charge density file is still quite a big file.

Swyx [00:35:04]: Yeah. Right. You hand wave a bit, but it’s good enough. This is, this is the highest order bit. Yeah.

Ekin Doğuş Çubuk [00:35:10]: I personally don’t know of a better method to predict the stability of a new material. It’s definitely not perfect, but it’s definitely better than other methods I can think of. And yeah, so going back to your question, the reason we use them is for certain things it’s easier to simulate than experiment, and they kind of complement each other well. And then we can do this in a loop. We feel like simulations will never be enough by themselves, but in the loop of simulations, AI, and experiments, I think we can make progress much faster than before.

When Simulations Fail: Microstructure and Calibration

Brandon [00:35:41]: One thing that simulations let you do is scale up much more quickly. Because now compute is, much easier to bring online

Ekin Doğuş Çubuk [00:35:48]: Yeah

Brandon [00:35:48]: Than bringing in bringing on new labs. How do you avoid. Or let’s say you scale up a bunch of DFT, how do you avoid in having your models over-index on that and still be grounded that the real world is ultimately the truth you care about? If you. You know, how do you sort of balance that?

Ekin Doğuş Çubuk [00:36:05]: Well, we scale up the lab as well.

Brandon [00:36:06]: Yeah, you scale up the lab.

Ekin Doğuş Çubuk [00:36:07]: Right, yes.

Brandon [00:36:07]: But you could. I could imagine scaling up DFT to millions, hundreds of millions given modern compute, whereas your lab, I would imagine, your hundreds, thousands a day? I don’t know. Something still fairly. Well, even in a really high-throughput case you’re still fairly limited by comparison, right?

Ekin Doğuş Çubuk [00:36:27]: Yeah. One thing is humans, but also LLMs, are kind of pretty aware of the limitations of DFT. For example, even if we can do, as you said, hundred million trials with DFT

Brandon [00:36:39]: Yeah

Ekin Doğuş Çubuk [00:36:39]: We could never actually try their microstructure. And microstructure is this idea that DFT usually simulates the perfect crystal, but crystals usually actually aren’t perfect, and their microstructure, which is the structure at a medium order, can really affect the properties. Other issues, of course, like some properties like superconducting temperature cannot be easily simulated by DFT. Again even if you do a hundred million trials with DFT, it won’t be enough. So we have to rely on heuristics, experiments, et cetera.

Liam Fedus [00:37:08]: And so there’s a huge filter from all those calculations to actually what gets executed in the lab.

Swyx [00:37:13]: Is that human-mediated or LLM or a mix?

Ekin Doğuş Çubuk [00:37:17]: Can be a mix.

Swyx [00:37:17]: Yeah.

Ekin Doğuş Çubuk [00:37:18]: For example, for a long time, Materials Project, which was the best open source DFT database, Use experimental calibration for DFT. So they would actually use experimental data to calibrate DFT. We also do this internally here. Whenever there’s a certain chemical system we’re interested in you get experiments, you get simulations, and you can calibrate them.

Why Inorganic Materials Differ From Biology

Brandon [00:37:37]: This is another thing. I’ve talked to a lot of biologist friends, and they are always very confused when you say materials, you can’t just do an XRD structure and actually know what the structure is. Because if you’re in the biology world, you can look at a crystal structure of a protein and typically more or less with sub A well order angstrom accuracy reconstructed. What is the difference between the materials XRD, X-ray disfor — X-ray diffu- no

Ekin Doğuş Çubuk [00:38:05]: Diffraction.

Brandon [00:38:06]: Diffraction, sorry. Experimental XRD for materials versus, let’s say, the biology world. What information do you lose, and why is this sort of like a lossy projection?

Ekin Doğuş Çubuk [00:38:16]: It’s hard not to sound crazy when answering this question. But do you not feel like, organic chemistry and biology almost stem from a lower VC dimension, like lower Kolmogorov complexity model? Where like, you can almost represent as a one D sequence, and it can almost be compiled. So there’s a sense in which it’s, simpler in the Kolmogorov complexity sense.

Swyx [00:38:38]: One D?

Ekin Doğuş Çubuk [00:38:40]: Like just DNA is just

Brandon [00:38:43]: Yeah.

Ekin Doğuş Çubuk [00:38:44]: Or RNA.

Brandon [00:38:44]: I think most people would argue it’s more two D, but

Ekin Doğuş Çubuk [00:38:47]: Okay, well, yeah. I’m not a biologist, so

Brandon [00:38:49]: Yeah.

Ekin Doğuş Çubuk [00:38:50]: I’m sure they’re right. But with inorganic chemistry, virtually it’s crazy. It’s clearly not coming from such a low Kolmogorov complexity model because that three D inorganic crystals can host metals, insulators, superconductors, diamond. It’s really crazy different. People tried, but we could never find a one D representation. It’s very hard to find a SMILES string for three D inorganic crystals. And the way the atoms interact with each other is quite strange, like covalent bonds, ionic bonds, metallic bonds. But yeah, this is, it’s very interesting. I wonder sometimes have you ever looked into issues with solid electrolytes? Why people cannot replace liquid electrolytes in batteries with solids? There are all these dendrites that form. Basically the lithium will form these structures that go into a solid electrolyte and crack it. And you look at that, and you think this is so primitive compared to biology. In biology, these systems can do such complex nano technology- technological progress, whereas we can’t even make two interfaces work with each other because lithium will break it. Yeah, but then I guess inorganic stuff can be very resistant to high temperature. You can make space shuttles. You can make silicon, compute Moore’s law. So yeah, it’s a trade-off.

Brandon [00:40:05]: Yeah, we’ve talked about this with several other guests. I think one of the big differences in my mind is that biology, we have a toolkit that you can basically borrow from millions of years of evolution

Ekin Doğuş Çubuk [00:40:19]: Yeah

Brandon [00:40:19]: That has already given us all the tools, and we can just reproduce those. But also evolution also constrains the yeah, the VC dimension to be fairly low for proteins and so on. The emergent beyond- behavior of large-scale systems like cells can be much more complicated.

Ekin Doğuş Çubuk [00:40:36]: Yeah, and consciousness.

Brandon [00:40:37]: Yeah. Yeah, but maybe even just more specifically for. If you just look at an X-ray, you had a perfect crystal, and you shoot an XRD at it, and you get the you get the spectrum. That still doesn’t actually tell you uniquely what the crystal structure is, right?

Swyx [00:40:54]: No.

Brandon [00:40:55]: So, and that’s something that

Swyx [00:40:56]: What?

Ekin Doğuş Çubuk [00:40:57]: It’s just average.

Brandon [00:40:58]: Yeah.

Ekin Doğuş Çubuk [00:40:58]: It’s just average behavior. But like, for example, disproportionation is a very complicated thing, right, where it’s possible that you have the perfect crystal, and on average, it looks like this perfect XRD. But in reality, there is some other pattern going on where atoms slightly shift to the right or left, but on average, they’re in the middle. Yeah, it’s a difficult life, yeah.

Brandon [00:41:16]: Whereas with proteins, you basically know there’s some sort of underlying protein follows some rules, so you can, you can combine a forward model with that.

Liam Fedus [00:41:23]: Right. And then you can kind of move and move towards single crystal XRD, and that can help.

Swyx [00:41:29]: When we started the Science Pod, there were all these elements of AI in science, AI in math, AI in physics, and all these things. And we there was this kind of pissing contest which is harder. And I think you can make the scientific argument that materials science is the hardest.

Liam Fedus [00:41:43]: Well, we didn’t pick it necessarily for hardness. There’s actually a lot of areas where it is easier. We have these great simulators for large classes of materials, whereas in biology, it’s incredibly difficult to model a cell or an organ or a full organism. So that’s a huge advantage.

Swyx [00:42:01]: Yeah. That, isn’t that interesting that it, at the smallest level, it is one-dimensional or two-dimensional, but and then it just has many orders more scale and structure, and that actually introduces the complexity. It’s, that’s weird. Whereas maybe complexity in materials is more the microstructure.

Liam Fedus [00:42:20]: Yeah, I still — you still get different complexity at the different, length scales for materials as well. So yeah, microstructure and beyond.

Dream Materials: Superconductors, Magnets, and Batteries

Brandon [00:42:26]: Yeah.

Swyx [00:42:27]: Yeah. Just sort of a brief sort of palate cleanser. In terms of science fiction, let’s say you can invent whatever materials that you want. What is more valuable? You know, I have, I have a list here. There’s obviously room temperature super- superconductors. But also, you mentioned batteries. I’ve always thought batteries. Just do better than lithium-ion, you’re good. That’s stood around for a hundred years. And then the last one is carbon nanotubes, mostly for the space elevators. I don’t know if, like. Those are, those are the three that come to mind. Are there any that people talk about in your world that are the dream?

Ekin Doğuş Çubuk [00:43:03]: Definitely superconductors

Brandon [00:43:05]: Yes

Ekin Doğuş Çubuk [00:43:05]: Magnets.

Brandon [00:43:06]: Right.

Ekin Doğuş Çubuk [00:43:06]: Especially if we can lower the dependence on rare earth or transition metals that are hard to source, like cobalt, reducing cobalt in batteries. One of the things I think is very fundamentally important is, if we can get close to Landauer limit for compute energy efficiency.

The Landauer Limit, Reversible Computing, and Energy

Brandon [00:43:24]: Can you, explain Landauer limit?

Ekin Doğuş Çubuk [00:43:26]: Yeah. So if you look at how much energy we spend per, like FLOP or compute, it has followed the Moore’s Law type behavior, but it’s gotten exponentially more efficient. So we’ve been spending exponentially less energy per FLOP or compute. I haven’t checked recently. This was the case ten years ago when I last

Brandon [00:43:44]: Yeah, that’s great. I remember when I first learned about it, looked at that, and we were, what ten, twenty orders of magnitude away from it or something.

Liam Fedus [00:43:50]: Beyond the limit, yes, right.

Brandon [00:43:50]: And the last time I looked, it’s. We’re weirdly close. We’re within a factor of I don’t know three or four orders of magnitude away.

Ekin Doğuş Çubuk [00:43:59]: Yeah.

Brandon [00:43:59]: It’s actually like

Liam Fedus [00:44:00]: I just exponentially rolled out over many years.

Brandon [00:44:01]: Yeah, exponentially over many years are

Ekin Doğuş Çubuk [00:44:03]: Four or five, yeah.

Brandon [00:44:04]: Yeah, pretty impressive.

Swyx [00:44:05]: But in your lifetime, you

Brandon [00:44:06]: Yeah, no I. Yeah, since, just since I discovered this, what, ten or fifteen years ago? Yeah, no.

Ekin Doğuş Çubuk [00:44:11]: And one of the things we have to do is dissipate heat, because as these, computations are happening, it’s producing heat, which has to happen, but then we have to dissipate it. So if we can get close to the Landauer limit for how efficiently we do computation, that’s amazing for humanity, right? Because that means now we’re doing computation, which is probably one of the most fundamental things we do as humanity, but as efficiently as possible from energy perspective, which is a real occurrence in the universe.

Swyx [00:44:34]: That get affected by quantum computing? I’m just going to throw it out there. Theoretically, massively, embarrassingly parallel compute.

Ekin Doğuş Çubuk [00:44:41]: So it definitely gets affected by reversible computing. But honestly, I’m not an expert. I don’t understand. So if you can do reversible computing now, you’re not. You don’t actually have to spend energy to do computation because you’re not actually deleting any information. It’s reversible. But I don’t know if that works. I don’t know. Quantum computing, I assume, has to obey these laws somehow because they are still, like. They have to obey thermodynamic

Brandon [00:45:06]: It’s unitary, so

Ekin Doğuş Çubuk [00:45:07]: Yeah, exactly

Brandon [00:45:07]: It’s reversible, so for now.

Swyx [00:45:10]: One of my most memorable conversations with Elad Gil was he actually was a very skeptical person about quantum computing.

Ekin Doğuş Çubuk [00:45:15]: I see.

Swyx [00:45:15]: He’s like, he’s like, “Even if you had it today, there’s no applications.” I’m like, “Whoa.”

Ekin Doğuş Çubuk [00:45:19]: I

Swyx [00:45:19]: It breaks, it breaks the it breaks the security RSA. That’s it.

Ekin Doğuş Çubuk [00:45:22]: I agree with that. Like for them, people talk about how if you had a quantum computer today, you could simulate things so much better. And I asked them, “Okay, let’s accept that. What would you simulate?” Like, you’re still simulating perfect crystal. You still have the issues that DFT has if DFT was perfect in its prediction ability. So yeah, I do think people are kind of glossing over some things because quantum computing is so exciting. If we can compute in a quantum logic space instead of classical logic, it’s just so exciting that people are glossing over what it would actually do when it’s made. Maybe that’s fine because maybe once it’s made, it will do amazing things we can’t even imagine.

Putting Intelligence Into Every Laboratory Instrument

Swyx [00:45:57]: Good. Okay. I was going to move to the automating the lab side, where you’ve closed the loop on your experimentation. You can talk about robotic arms, talk about. But it’s basically just everything you’ve done here. My favorite quote from you was that every piece of equipment in your lab is going to have a hundred and forty IQ. Okay, what does that mean?

Liam Fedus [00:46:17]: In the early days, as we were scaling things up, we realized that we had huge bottlenecks imposed just by operating machinery. So, for one set of machinery, we would have technicians and scientists looking for particular morpho- morphology and trying to see, okay, what actually were we making, see if this is consistent with our intentions. And we really quickly, as we scaled up the lab, came into these bottlenecks where it just wasn’t keeping up. So we started doing some programmatic approaches to capturing the data on SEM. And the data that was being surfaced from this very simple program where it sort of takes a field of view, captures, zooms in captures, was just not all that useful. It was a little too dumb for what we really wanted to get. And so at that point, it was then, really pertinent to build AI systems directly onto the machines to start controlling these things. And now they have the full context as to what we were trying to achieve. So what was the intent of the experiment? What were we trying to synthesize? What were some other experimental evidence? And it’s actually looking, in the machine and capturing that data. And this is really valuable now because we’ve been talking about these hidden variables not being able to capture everything, but if you can do more intelligent data capture at the time of that experiment, your data for future AI systems and future computational predictions is that much better. You’ll have just a richer set of data. And so that’s sort of what we mean by just, everyth- everything on the lab has to be incredibly intelligent to just make the data as useful as possible. So that was, that was sort of the motivation and inspiration behind that.

Swyx [00:47:57]: What’s the state of the art there in terms of putting intelligence on every device, right?

Liam Fedus [00:48:02]: Well, I think one interesting aspect, is what’s the latency

Swyx [00:48:08]: Exactly

Liam Fedus [00:48:09]: Of controlling that instrument?

Swyx [00:48:10]: Because there’s cloud latency, but then there’s also device compute, which

Liam Fedus [00:48:13]: That’s right

Swyx [00:48:13]: Also.

Liam Fedus [00:48:14]: And then there’s sort of a time scale associated with different physical processes. And if calling out to an API or something is simply too slow, so the amount of reasoning or tokens or tool calls, is just not matched to the latency of that actual process, then that’s infeasible.

Swyx [00:48:33]: Okay. That is right.

Liam Fedus [00:48:34]: It’s almost like, self-driving cars, right? So.

Brandon [00:48:37]: I. Real quick. I just am actually surprised to hear that because the latency I’d imagine for reasoning seems small compared to a lot of these things take hours to run, right? Or maybe, you know. Or maybe I. Maybe your experiments are much faster or high-throughput or something. I’m just surprised to hear that, actually.

Liam Fedus [00:48:52]: So ultimately, the experiments do take, many hours, days, but there could be particular steps where latency would matter.

Ekin Doğuş Çubuk [00:49:00]: Yeah. You might want to have finer control. Another thing is, how expensive it is because the same reason some of these models can be really slow in analyzing data also makes them …

Watch the full show on YouTube: https://www.youtube.com/watch?v=MWX36ZYnsm0

Last call for regular tickets for AI Engineer NYC! As an exclusive for Latent Space subscribers, the first 30 of you can take a 30% off code if it helps - for new tickets only, no refunds! See you in 2 weeks!

While we tend to cover industry on the pod, every so often we celebrate a clearly emerging superstar PhD. In 2024 we featured Shunyu Yao, who went on to build Operator at OpenAI and is now Chief AI Scientist of Tencent. In 2025 we featured Jack Morris, who went on to cofound Engram at $600m and is now a leading voice on continual learning.

This year we are proud to feature the work of Alex Zhang of MIT.

From GPU kernels and KernelBench to Recursive Language Models, Mismanaged Geniuses, and massive multi-agent swarms, Alex Zhang is exploring how much capability we’re leaving on the table by wrapping increasingly powerful models in primitive systems.

RLMs took over the timeline early this year:

and an RLM based harness was the first to ~solve ARC-AGI-3 before OpenAI’s Astra:

and is even today, influencing new research that has more extreme implications than RLMs:

We go deep on GPU Mode and AI-written kernels, research taste and why academics should take bets industry labs won’t, GEV and alternatives to the standard autoregressive language model, and the idea of harnesses as compositional generalizers. Alex explains RLMs, context offloading, programmatic subagent calling, Prime Agent, persistent subagents, and why the “language model” of the future may actually be an invisible swarm of agents underneath a simple interface. We also discuss OpenAI’s massive agent experiments, Kimi swarms, open-ended research at Sakana AI, speculative programmatic tool calling, capability overhang, Neuralese, and where Alex thinks the next big research opportunities may lie.

We discuss:

* Why AI-generated GPU kernels still leave substantial room for human expertise

* How one expert insight can potentially replace enormous amounts of brute-force token search

* Why PhD students should take research bets that initially look trivial, weird, or pointless

* What SWE-bench, RLMs, ReAct, and Quiet-STaR reveal about research taste

* GEV and why a language model does not have to mean an autoregressive text-to-text decoder

* Why Claude Code, Codex, and Pi are structurally more similar than they look

* How harness design can improve compositional generalization across tasks and domains

* RLMs: context offloading, code execution, recursive subagents, and shared memory

* Prime Agent, continual harnesses, and persistent agent-to-agent communication

* Why the model you query in the future may secretly be an entire swarm or scaffold

* OpenAI’s 10,000-agent experiment, 130B output tokens, and ~$40M-equivalent problem solving

* Why much of an agent swarm may be wasted search — and why convergence is still hard

* Kimi versus OpenAI and different approaches to multi-agent systems

* Open-endedness, Sakana AI, and finding hidden gems in enormous amounts of generated work

* Why current frontier models may already have a large capability overhang

* Speculative programmatic tool calling and overlapping tool execution with generation

* Whether English, code, or an entirely new “Neuralese” constrains how models reason

* AI for science, fast-moving benchmarks, and how Alex chooses what research problems to bet on

Alex Zhang

* Website: alexzhang13.github.io

* X: @a1zhang

Timestamps

00:00:00 Introduction

00:00:49 GPU Mode, KernelBench, and AI-Written Kernels

00:07:38 Human Expertise vs. Brute-Force AI Search

00:13:20 Research Taste and Taking Big Bets

00:19:28 GEV and Rethinking the Language Model

00:29:03 Video Game Agents and the Harness Problem

00:31:01 Why Claude Code, Codex, and Pi Are So Similar

00:36:42 Harnesses as Compositional Generalizers

00:44:24 RLMs Explained

00:52:01 Prime Agent and Persistent Subagents

00:57:41 RLMs in the Wild

01:00:30 OpenAI Swarms and the Future of Language Models

01:07:26 Open-Endedness and Sakana AI

01:15:52 Kimi vs. OpenAI Agent Swarms

01:20:06 Capability Overhang and Speculative Tool Calling

01:28:19 Neuralese, Future Research, and AI for Science

Transcript

Introduction: Alex Zhang, RLMs, and GPU Mode

Swyx [00:00:00]: All right, we’re here in the studio with Alex Zhang, I guess most famously of, RLMs, but you have a few other affiliations. Welcome to the show.

Alex Zhang [00:00:12]: Yeah. Thank you for having me.

Swyx [00:00:13]: Yeah. I guess GPU Mode as well?

Alex Zhang [00:00:15]: Yes, GPU Mode as well.

Swyx [00:00:16]: You were shepherded in by Mark Saroufim. Not everyone gets that kind of welcome.

Alex Zhang [00:00:19]: Yep. Yeah. Yeah. I’m very close to all the people in GPU Mode, so yeah

Swyx [00:00:23]: Yeah

Alex Zhang [00:00:23]: We often end up working together in various capacities, like even beyond just GPU Mode itself, so.

Swyx [00:00:29]: Yeah. Can we explain, so people who are not that close Don’t know about this. It’s just a-- it’s a Discord. It used to be focused on, I guess, CUDA Mode, and then generalized a little bit. it was started by Mark.

Alex Zhang [00:00:41]: Yep.

Swyx [00:00:42]: It was basically like. To me, it’s like the hiring pipeline of the PyTorch team.

Swyx [00:00:45]: And then you left PyTorch.

Alex Zhang [00:00:47]: Yep.

Alex Zhang [00:00:49]: Yeah. So it used to be, I think, it started actually around when I was in college, in like 2023. I think it was started by Mark, Andreas, and Jeremy Howard. The original premise was just like, it was a GPU-- or it was a Discord dedicated to learning how to write GPU kernels, and they had, like, lectures. That was basically the extent of it. and I got interested in it because I was writing GPU kernels. It was actually out of, like, pure chance. I was interning at Snapchat at the time, and

From CUDA Mode to GPU Mode

Swyx [00:01:22]: Rexis.

Alex Zhang [00:01:23]: Yeah, I was very bored with Rexis. So, they had a project where, like, they were interested in writing. It was this paper called Infinite Attention. It was like a Google paper.

Swyx [00:01:35]: Yes, we’ve covered it on Paper Club.

Alex Zhang [00:01:36]: Yes, yeah. So I was interested in whether or not you could write specialized kernels for it at Snapchat. it didn’t. Nothing really came of it, but I joined GPU Mode. At the time, it was called CUDA Mode, I think for, like, legal reasons or something, they changed the name. But I met Mark, I met Matei, I met a bunch of other people that were very involved in the community. And then Mark had pitched this idea called Popcorn, which was now what you see as the leaderboard today. But the general idea was like. I think all of us had this, like, intuition that GPU programming is, like, very similar to if you guys have done, like, competitive programming. It’s a, it’s a not. I don’t mean to say, like, they’re transferable skills.

Popcorn, KernelBench, and Automating GPU Kernels

Swyx [00:02:17]: You have constraints. You code golf a little bit.

Alex Zhang [00:02:19]: Yep.

Swyx [00:02:19]: Yeah.

Alex Zhang [00:02:19]: Yeah. And there’s like. There’s actually a surprisingly small space of optimizations that people do. and there’s actually not that many kernels per se that people are interested in optimizing. And so we kind of had this thought that, like, if you had enough data, like, in the same way that Codeforces, there’s millions of problems. If you could do this with GPU code, like, you could scale and automate kind of GPU kernel development, which for researchers is a huge deal. ‘Cause I think one of the bigger bottlenecks. Like, if you look at, like, Mamba, for example, like, they release the paper with kernels because otherwise, like, you can’t really use it in any meaningful way? And not everyone has, like, a Tri Dao on their team. So we’re very interested in this. KernelBench kind of spawned from that too, of like, can we get LLMs to automate, GPU kernel code? And I think that was like a. It was a very fun time. It was like between college and my PhD, and yeah, I had a really pleasant time doing stuff with GPU Mode. Now I kind of just help with the lectures sometimes. I’m not as involved, and I think in general, like, we don’t have as many competitions as we used to. But, yeah, I still keep in touch a lot with everyone there.

Swyx [00:03:27]: Is there a friendly rivalry? Because I think the previous community that used to do this was like MLSys, MLPerf

Alex Zhang [00:03:32]: Yeah

Swyx [00:03:32]: Kind of thing. Is there a friendly rivalry? Is this like just new generation MLPerf, or what’s going on?

Alex Zhang [00:03:38]: The nice thing about GPU Mode is that it is also a community in the sense that, like, a lot of the lectures are very easy enough for a beginner to follow and ask questions and things like that. And, like, the competitions are, like, somewhat, not secondary, but, like, you can participate in them to learn. I think with a lot of, like, MLSys, MLPerf kind of benchmarks, like, for the most part, like, only serious labs and companies participate. Like, seriously in them, at least. That was my understanding of it. I could be wrong. But I think also beyond GPU Mode now, one thing that has been really exciting is there’s a lot more websites and, like, people that work on hosting competitions. Like, I think there’s this.

Alex Zhang [00:04:21]: I think there’s this website called, like, LeetGPU or something, and it’s like leet code for GPU problems.

Swyx [00:04:26]: Wow.

Alex Zhang [00:04:26]: There’s, like, other ones that I’ve. Like, we’ve, we’ve seen. Like, there’s many that have kind of spawned and, like, talked on GPU Mode, and like, it’s very exciting in the sense that I think GPU programming used to be super niche, like when I was interested in it. And the only reason I got interested in it was Tri Dao gave a talk at Princeton because he was, applying for faculty. he is faculty there now, but I listened to his talk on FlashAttention in like 2023, and I was like, “Wow, this is like the coolest thing ever.”

Alex Zhang [00:04:56]: And I was like, “This is like. This is what everyone should be working on.” I guess, like, vLLM and stuff had come out too, and it was like, “Oh, we should be writing kernels.” But now it’s like, everyone writes kernels. Like, everyone. It’s, it’s. I think it’s actually almost saturated in some sense, as a field.

AI-Written Kernels and the Verification Gap

Vibhu [00:05:11]: Any interesting takes for people that wanna get into it? So I think one of the biggest news is GPT-5.6 Wrote more efficient kernels, so Terra and, Luna could be 80% cheaper.

Alex Zhang [00:05:24]: Yeah.

Vibhu [00:05:25]: And then we’ve seen other competitions where people are, like, setting records, and they’re like, “We’re doing some auto research loop,” and these are people that don’t have a background

Alex Zhang [00:05:34]: Yep

Vibhu [00:05:34]: In any kernel writing, right?

Alex Zhang [00:05:36]: Yeah. So even on the GPU Mode leaderboard, if you look at, like, a lot of the recent problems, almost all the solutions are AI generated. However, you’ll notice on the leaderboard. So there’s this guy named Gauners who’s, like, a very, like, regular member of GPU Mode. We’ve always known for a long time that he’s, like, a super cracked, like, GPU kernel writer. One thing we discovered on this leaderboard is, like, almost ev-- Like, he also used AI to help him with these solutions, but for the most part, like, he helped prompt and move it in certain directions. we found that, like, his kernel was, like, basically the only one in, like, the top 10 that was actually stable in, like, actual, like. end-to-end systems. And it does bring into question, like, it’s not. Like, GPU kernels have a verification problem. Like, we’ve kind of known this. It’s been a problem since KernelBench was released. Like, there’s a lot of reward hacking that goes on. But you al-- Yeah, you also notice, like, the lines of code is a lot smaller, but

Vibhu [00:06:33]: Yeah, I was gonna ask, is that noticeable, or is it just

Alex Zhang [00:06:36]: Yeah, no, it’s, it’s

Vibhu [00:06:36]: Okay

Alex Zhang [00:06:36]: It’s definitely, like, very important, and I think, like, it’s, it’s really interesting that still there’s a lot of alpha in being good at writing GPU kernels.

Vibhu [00:06:44]: Okay, so there is a gap from verifying

Alex Zhang [00:06:45]: There definitely is, yeah. I think, like. And this applies to a lot of AI systems as well. Like, I think, even with the most recent, like, math proofs and stuff, like, it doesn’t necessarily mean mathematicians are obsolete. these companies still hire mathematicians, like, to do, whether it be, like, data labeling work or even just, like, steering the models to solve problems. Like, there is still a lot of alpha in being, like, knowledgeable in these things, so.

Swyx [00:07:12]: Is it just knowledge, or is it also there is just more planning, and is there, an emergent style of planning that works better?

Alex Zhang [00:07:22]: I think it’s, it’s a mix of. Maybe this is what you mean, like intuition for

Swyx [00:07:27]: Something like that

Alex Zhang [00:07:28]: How to solve the problems.

Swyx [00:07:29]: Like, for example, I always diagram my code.

Alex Zhang [00:07:31]: Yeah.

Swyx [00:07:31]: Right?

Alex Zhang [00:07:31]: Yeah.

Swyx [00:07:31]: And then, like, if there’s a part of the diagram I don’t understand, I work until I understand it. Otherwise, I, it’s not allowed.

Alex Zhang [00:07:37]: Yeah.

Swyx [00:07:38]: Yeah.

Alex Zhang [00:07:38]: So I think it’s, like, it’s a mix of those things of, like, the people who work. Like, the people who know how to look at these problems and how to solve them, like, also know how to use AI to do them. Because, like, you’re acting as a very strong verifier. Like, if you are knowled- or if what to do and you are also. Like, I think the thing that we’ve kind of discovered with all these agent swarms and things like this is, like, when you throw enough compute at a problem, you, like, can sufficiently explore solutions to that problem. But oftentimes, like, maybe you can burn, like, 100 billion or a trillion tokens on something, but if you bring in someone who knows something about the problem, they can uncover something for the model that would, like, erase that one trillion token spend. it’s not, it’s not super clear, like, what exactly the trends are here. But I think, like, there are so many problems in the wild still right now that we want to solve, and, like, we can’t afford to just always, throw as much compute as possible at it. Like, there is still an efficiency aspect of all of these things that is super important.

Speed-of-Light Limits, Memory, and Megakernels

Swyx [00:08:41]: Is there, like, a theoretical right answer that you just calculate based on physics, and then you just get close to the physics limit?

Alex Zhang [00:08:49]: Yes. So for GPU kernels, you can compute. It’s actually not that easy to compute sometimes, like, depending on how complex the problem is. Like, for matrix multiplication, it’s very easy to compute, this, like, speed-of-light kind of, estimate of what the fastest kernel can be. And, like, this is also assuming, like, maybe all of your, all your data starts on the CPU, or maybe it starts in DRAM, on the GPU, et cetera. Like, this changes these numbers slightly, but

Swyx [00:09:18]: The transfers and all these things, yeah.

Alex Zhang [00:09:20]: Yeah. But I will say, like, it’s not clear, though, like, in a lot of cases if it’s even possible to hit this theoretical number, if that makes sense. Like, this is assuming, like, perfect overlapping and transfer of data, and, like, there’s maybe some bottleneck that you can’t get around. But often, the kernels are not even close. Like, that we write are not nearly close enough to this number to be, like, meaningful at all.

Swyx [00:09:43]: Yeah. And is it speed that matters? Do you also care about, obviously memory, which

Alex Zhang [00:09:48]: Mm

Swyx [00:09:48]: Feeds into speed? Do you care about power consumption? So one of my, one of our top pods of the year was Geoff Dean, who was like, “Actually, I just tracked the microjoules or, like, the nanojoules, picojoules.”

Alex Zhang [00:09:59]: Yeah, it’s often picojoules today.

Swyx [00:10:01]: Picojoules.

Alex Zhang [00:10:01]: Yeah.

Swyx [00:10:01]: Do you care about that?

Alex Zhang [00:10:03]: So I don’t.

Alex Zhang [00:10:04]: Yeah. I guess maybe I’m not, I’m not as

Swyx [00:10:06]: But everything here is speed, right? Like

Alex Zhang [00:10:07]: Everything here is speed

Swyx [00:10:08]: Nobody’s counting picojoules.

Alex Zhang [00:10:09]: But I-- There’s a caveat here, which is, I think, like, there is speed in the context of a single kernel, and there is speed in the context of a larger problem, like maybe the N10 model. Because, like, one thing to consider, and this is why it’s important to talk about what speed-of-light is referring to, because in these cases for the kernels, like, we always start with everything in, like HBM, for example, right? But you can imagine that, like, an end-to-end like an end-to-end model, what you might wanna do between two layers is, like, you might sacrifice the speed of the first operation to keep things in the cache for the second operation. And, like, these are things that, like, you can’t really get out of, in isolation with, like, these kinds of kernels. And, people call this, like, the fusion or, like, the

Swyx [00:11:01]: Megakernel

Alex Zhang [00:11:02]: Kernel fusion problem. Yeah, or, like, megakernel stuff. And it generally only applies, like, when you are, like, memory-bound in most cases. But this is something that, like, also there is this question of, like, as these models get better, like, should we just be generating like, megakernels? Is that, like, what we want?

Vibhu [00:11:20]: What’s, what’s your take?

Alex Zhang [00:11:21]: I think that this is really difficult because you need the data to do this. And I think, like, I have yet to see an example in the wild of, like, we bootstrap the ability to solve a very difficult class of problems without any examples. and I think, like, the other reason why I think maybe this isn’t that interesting is that at the level of an individual kernel, A, like, they’re not that, they’re not as complex, but B, you’re somewhat confident that there’s not as much structure in a single kernel. But, like, in a megakernel, like, I would be more inclined to believe that, like, a compiler would be better here. Like, some compiler over, like, higher level- Ops makes sense, because in general, like actually, I think mega kernels are very, like the pieces are very composable of like the individual kernels. There’s some areas where you might wanna do like weird fusions and everything, but in general, I think these are cases that like a compiler can probably handle. And there is a company that’s working on this from what I understand that has given some talks on GPU mode as well.

Swyx [00:12:28]: Yeah, I wanna basically cluster all the GPU mode discussions here because obviously there’s other parts

Alex Zhang [00:12:32]: Right.

Swyx [00:12:32]: That we need to move on to.

Vibhu [00:12:33]: I think there is something to plug. You guys do host a lot of really good lectures. They’re all on YouTube. People can follow. And you

Alex Zhang [00:12:39]: Yes

Vibhu [00:12:39]: Lead quite a bit of it. You’re still quite involved.

Alex Zhang [00:12:41]: I used to. sometimes I still do. I think they’re mostly Mark. Mark is the one who usually does them. Matei does sometimes as well, but, yeah, I highly recommend them. They are extremely good resources. Like, I think it’s kind of crazy how much people share on there, so yeah.

Swyx [00:12:59]: ‘Cause like if you’re there, like you’re very, like you’re exactly the right audience?

Alex Zhang [00:13:03]: Yes, exactly.

Swyx [00:13:03]: Like this isn’t gonna reach the mainstream.

Alex Zhang [00:13:04]: And there’s a lot of like introductory material as well, that we’ve put on, that I think is useful for people.

Benchmarks, Princeton, and Research Taste

Swyx [00:13:10]: KernelBench was kind of influential. I just wanna see like, that was last year.

Alex Zhang [00:13:13]: Yep.

Swyx [00:13:14]: What other ongoing work do you wanna shout out that people should pay attention to? ‘Cause obviously you’re involved in this

Alex Zhang [00:13:20]: Yep

Swyx [00:13:20]: Field.

Alex Zhang [00:13:20]: Yeah. I will give maybe the background story of like I am actually involved in a lot of benchmarks, or I used to be, maybe prior to my PhD. It started because I was at Princeton. I worked with the SWE-bench team there.

Swyx [00:13:33]: John, Carlos

Alex Zhang [00:13:33]: John, Carlos

Swyx [00:13:34]: Ofir

Alex Zhang [00:13:34]: And Ofir. They’re all great. Like

Swyx [00:13:36]: Karthik

Alex Zhang [00:13:36]: I love them. Yeah.

Swyx [00:13:37]: There’s basically this, like I think people don’t understand how much benchmarks come from the same group

Alex Zhang [00:13:42]: Yeah

Swyx [00:13:42]: At Princeton.

Alex Zhang [00:13:44]: It is crazy.

Swyx [00:13:45]: Do Xun Yu?

Alex Zhang [00:13:45]: Yes. Yeah.

Swyx [00:13:46]: We had him on a pod before. Now he’s like running Tencent.

Alex Zhang [00:13:48]: Yeah, now he’s like, he’s like a superstar. when I met him, so he was advising my friend Michael Tang, who is now at Anthropic, but they worked together a lot. We were like the two undergrads in Karthik’s lab. I-- And then some others joined later as well. But yeah, Xun Yu is great. I did not know he was like such a superstar until like later on, like after I left, but

Swyx [00:14:12]: Yeah. like, okay, so there are very few PhD students. Like yours is like the next one. Like once a year, we feature someone like who is like basically entire PhD, has been like on target.

Swyx [00:14:25]: There’s not that many of them. Xun Yu was like clearly one of them. And, Jack Morris is another one. And like, we talked, before the show, we talked about research taste.

Swyx [00:14:33]: Right? Like somehow some grad students just have a very blessed career where like, yeah, mostly like, yep, this is like going to stick around, relevant, everyone should know this.

Alex Zhang [00:14:42]: Yeah.

Swyx [00:14:42]: And then others, just nothing.

Alex Zhang [00:14:44]: I think this is also true of like even people within like industry labs as well. I think it’s just like grad students are a lot more visible. So you just see, like you see, like, there are some people who

Swyx [00:14:56]: Yeah, you can publish

Alex Zhang [00:14:57]: Who really like get lucky and like, or it’s, it’s a mix of being lucky and also being very smart and things like that. I think like with research taste as well, like I think it gets developed through opportunities, at least in my case. Like I got-- I was very fortunate to have like taken the path that I took, like working at Princeton and then like later, like finding my like Omar at MIT. Like he’s a fantastic advisor. I will say, though, I find that the most successful research from grad students or like in academia comes when people care about problems that maybe like most people in industry are not looking at. I think this is the issue that like a lot of grad students work on things that benefit, like that look good to an industry lab. Like for example, they’ll work on some, like some benchmark that’s really popular now. I think benchmarks in 2023 were a very different story than benchmarks now. There are a lot of people that work on like harnesses and like meta-harnesses and like specific harnesses for XYZ task. And when you really think about it, the reason someone would work on this is like maybe there’s like a clear goal shaped around the models that we have today of like, this is what I want to see. But like, I’ll give, I’ll give the like RLM, like the recursive language model paper as an example, because I think like it’s a super simple idea. I think when it came out as well, like there were a lot of people that were like, when they see something like that, they’re like, “What is even the purpose of this?”

Swyx [00:16:27]: Or like too cool.

Alex Zhang [00:16:28]: Yeah. Like why, like what? This is just subagents or something, right?

Alex Zhang [00:16:32]: And I think it’s like when you get a reaction like that, it’s almost like a good sign in the sense that like it’s clear that people aren’t thinking about what the purpose of this is. And I will give another example of like SWE-bench. When SWE-bench came out, Ofir loves to tell this story. When it came out, like nobody cared. Like everybody was like, “This is an impossible task. Like why would we ever even consider this as a benchmark?” And it wasn’t until Devin came out that everyone was like, “Whoa, like this is something we wanna hill climb.” And I think this is, this rings true for. You tend to see that a lot of ideas. I think like the. My favorite, I guess, example of this is Eric Seligman’s work, with like STaR and like Quiet-STaR. Like I think when you read the paper, at least when I first read the paper, I was like, “Is this not like an obvious idea?” Or maybe not. I don’t know. I was like, “Oh, this seems really simple.” Or like chain of thought, and the same thing. Or like Xun Yu’s react. It’s like, okay, like, yeah, sure. But then like when you really think about it’s like why. What is the value of the paper? And I think it comes from like, it tells a bit of a story as to like what you want the field to look like. And that is something that it’s very hard to do this in academia because if you look at all these papers, Quiet-STaR, ReAct, RLMs, SWE-bench, none of these papers are. It’s not like a GPT-6 Astro release? It’s not like everyone’s like, “Oh my gosh, like I’m gonna use this now and this is the best thing in the world.” Like academia just can’t afford to do this, at least right now. I. There’s a whole slew of reasons why I think that should change, but I think it’s like. If you don’t have. As a PhD student, I think you’re in such a unique position where you can work on literally whatever you want for the most part. If you’re not taking advantage of that and working on things that, like, nobody cares about or, like, people see as some trivial thing, like, “Oh, I thought about this, but, like, I don’t use it,” I just think, like, in the end, the research is just never gonna be that interesting because you kind of need to take big bets if you’re gonna be in academia. Because otherwise, I think, like, just go to an industry lab. Like, they have tons of resources, tons of talent. Why constrain yourself in an area where you don’t have a lot of resources and, like, there’s not even that many people around? And I think it’s just. it literally just comes down to, like, big bets. like, you just have to take big bets, and, like, a lot of them will fail? Like, that’s just. it’s, it’s natural. But I think

Alex Zhang [00:18:55]: That is, as a PhD student, like, that’s the biggest advantage you have over any single person at another lab because you don’t have to deal with bureaucracy and all these other things.

Swyx [00:19:07]: Fair enough.

Alex Zhang [00:19:07]: Yeah.

Swyx [00:19:08]: I ask a lot of people this question, and usually they hand-wave away. So I think. I appreciate that you’re actually giving a thoughtful response on Like, no, like, this is your unfair advantage because everything else is biased against you, basically.

Jev and Breaking the Autoregressive Decoder Paradigm

Alex Zhang [00:19:20]: Yeah, exactly. And so, like, it’s honestly. I will bring up Jev as an example because it’s, it’s not an academic

Swyx [00:19:26]: Wow, okay.

Alex Zhang [00:19:27]: It’s not an academic project.

Swyx [00:19:28]: Yes.

Alex Zhang [00:19:28]: I want to bring this up because this also happened with RLMs and, it happens with many other works. Like, things get overhyped, right? To an extent, like, something gets overhyped and then people are like, “Why is this overhyped?” Like, “This is trivial. This is stupid.” And I saw the same thing with Jev because I think the release was like. there is this whole thing about, like, academics, or they’re not an academic group, but, like, people have to do branding and they have to, like, kind of market their research. And so, like, I understand, but I think there was a lot of discourse about Jev just being, like, something we’ve known for years. And I think it’s kind of missing the point of, like, why is such a system so interesting? It’s why is it not just some stupid NLP classifier that, like, we’ve, we’ve been doing, back in our intro ML classes or something? I think what’s really interesting about Jev is that it kind of opens up this question of, are language models correct? Like, in the form that they’re in, can we consider a different design space other than text-to-text? Because what they’re doing is they’re basically saying like, “I will take advantage of this language model backbone. Like, I know it captures a lot of information about language, but I’m going to change the output space of the model to give you a trade-off, which is I will do very fast inference over.” Like, if you have some prior about this problem, like, let’s say I only need to make a binary classification. Am I gonna ask my language model to do this and pay, like, a 400X cost? Like, no. That’s-- it’s, like, silly, right? and I think for the longest time, because the labs are the only places that control, you’re never gonna use something other than, like, GPT-4 or GPT-6 or Fable because they’re the best models. But because of that, like, people have gotten kind of accustomed to this idea that a language model is just a autoregressive decoder. Like, we have accepted this. And I think when RLMs came out, it was the same thing. Like, one of the comments, like a very frequent criticism I got was like, “This is not a language model.” Or like, “When I look at this, like, I thought it was a new architecture, but it’s actually not.” And my response to that is like, “Well, a language model is just modeling language. It doesn’t have to be this transformer decoder,”? and Jev is really interesting in that, like, we now have a new

Alex Zhang [00:21:49]: Thing to tune, which is like, what is the output space and how does this affect inference latency? and I think we can actually start asking this about various parts of the language model itself. we are seeing this too with, like, loop transformers. It’s a similar idea of a lot of the attention around it was like, “This is a silly idea.” Like, “Why? Who cares about this?” But it’s like, it is a simple idea, but it’s actually. it opens up a whole new set of questions that I think, like, especially if you’re a PhD student, these are the things that you wanna answer. Because I think it’s like we don’t know. For Jev, for example, we don’t know how far we can take this. for loop transformers, we also don’t know how far we can take this. What if you loop only a subset of the model? what if you route to, like, only. like you have some router to different parts of the model? Like, can you mimic what you do in a harness inside of the model? What can you bridge between a harness choice and a model architecture choice? These are all questions I think that open up with works like this, and that’s, like, really exciting. Jev in particular, when I saw it, I was like, “This is actually really useful for RLMs.” Like, I think it’s, it’s. it makes sense ‘cause the biggest bottleneck in RLMs or swarms or systems like these is they’re slow. When you do multiple language model calls all the time, you’re not distributing your compute correctly because, like, maybe there’s something trivial that you just want a simple model to do, but you can’t do it because your language model is just this bulky thing? So I’m very excited. I think we will start to see new types of models emerge beyond just the bog-standard frontier model, and that is like. there’s so many things that you can do with these, like, new trade-offs.

Swyx [00:23:34]: I’ll also shout out Thinky with their interaction models.

Alex Zhang [00:23:36]: Yes. Yeah. Another great example.

Swyx [00:23:38]: Yeah. So, like, basically try to break the paradigm from sequence to sequence, decoder only, and just literally do anything else.

Alex Zhang [00:23:46]: Yeah.

Vibhu [00:23:46]: I think there’s, there’s a level of if you’re trying to compete, you’re not gonna compete with a Frontier lab doing an autoregressive

Alex Zhang [00:23:54]: No.

Vibhu [00:23:54]: Decoder on. Like, the amount of compute scaling resources they Even Thinky will not. okay, there may be one of the handful that can, but you’re not really gonna do much in that at least.

Alex Zhang [00:24:06]: I don’t know too much about Thinky, or I don’t wanna say anything either, but it’s like if their strategy is just to replicate OpenAI or Anthropic, like that’s a horrible strategy.

Alex Zhang [00:24:15]: Because, well, because, like, they just don’t have. Like, you kinda just have to think of it in terms of, like, what advantage do you have? And if you’re going to use the same setup. I’m sure they’re not, but it’s like if you’re going to do the same setup, like you’re basically competing on the things you can control, which is data and compute, and obviously they cannot compete with the Frontier Labs on that. So yeah, it makes sense that, like, if you are a neo lab, like. Actually, I don’t know if you would consider them to be a neo lab, but I guess, like

Vibhu [00:24:40]: Yeah. That’s why they’re, they’re in there.

Alex Zhang [00:24:42]: I guess they’re kind of a weird one, yeah.

Vibhu [00:24:43]: They’re in their list. They shipped Inkling. Like, they can’t.

Alex Zhang [00:24:46]: Anything other than OpenAI or Anthropic, maybe like Meta and GDM, like you just, you gotta do something else? Like, it just. It’s the sad reality, but I think. I actually think it’s a good thing. I’m very happy that, like, scale works and these companies will just keep doing it because it opens up, like, potentially new players, like if they uncover something really interesting. Because I sort of have my doubts that this is, like, seriously going on at Frontier Labs, ‘cause it’s like why would you do that? like, why would you take the risk of allocating a large amount of compute to new bets when the old bet already works? So

Vibhu [00:25:24]: And I think that’s what spins off a lot of neo labs, right?

Alex Zhang [00:25:27]: Yeah.

Vibhu [00:25:27]: You have a side bet and you don’t get compute, and you’re like

Alex Zhang [00:25:30]: Yeah

Vibhu [00:25:30]: “Okay, I’ll go, I’ll go do that.”

Alex Zhang [00:25:31]: Exactly.

Vibhu [00:25:31]: And, your example of the potential upside is something like Jev, which is X hundred times cheaper, comes out, and maybe it is language model.

Calibration, Fast Classification, and New Model Trade-offs

Alex Zhang [00:25:41]: Yeah.

Vibhu [00:25:41]: In this case, it’s just different.

Alex Zhang [00:25:42]: Yeah.

Swyx [00:25:43]: Yeah.

Swyx [00:25:44]: So no speculation on what Jev actually is?

Alex Zhang [00:25:46]: I guess I have some guesses for what it might be. I have seen some people say like, “Oh, it’s like a diffusion thing.” I guess that

Swyx [00:25:56]: Which is the parallel decode, right?

Alex Zhang [00:25:58]: Yeah, parallel decode. Honestly, I think regardless of what it actually is, ‘cause I think you can. I’ve seen some, like, open-source replications of it. What is really exciting to me about what they did is I’m not entirely sure what their optimization objective was and how they trained it. And I think, like, this is a thing for RLMs that we’ve also been thinking about, which is like, okay, like RLMs are a very simple idea. If I come out with this paper, like anyone can use it now. But what distinguishes The actual value of an RLM is whether or not you can train it properly, and whether or not maybe you can mold some architecture around the system to make it really good. And that’s something that, like, I’m actively working on, I guess. But I think for them, like, they figured out a way to train the system, which is completely non-trivial. Like, I actually don’t really know how they did it. And I’ve seen some comparisons online of, some people are claiming they used Qwen, or they post-trained on top of Qwen, but every open-source Qwen that you use is gonna be worse, ‘cause whatever they did to train it clearly works very well. And so that’s, that’s very exciting.

Swyx [00:27:04]: There’s one element of calibration Which, is a rare topic that I don’t think people even knew about or understood. We covered it with, our conversation with Clementine Foley of Hugging Face, and she used to run the evals, at Hugging Face, which is basically the idea that, models are attuned to give you the most likely next token. But, they’re gonna lie to you when you ask them, “How confident are you?” Because they’re just gonna give you the most likely next answer instead of, like, actually, like, no, let’s calibrate. Like, I am actually fifty percent sure, or I am twenty percent sure, and, like, let’s try to calibrate that. I would say, like, if anything, I think that actually that’s pretty easy to generate synthetic data around Because you can sort of see the truth and then synthetically generate a bunch of answers, have Jev classify it, and then compare with ground truth.

Alex Zhang [00:27:52]: Oh, I see.

Swyx [00:27:52]: That would be my reverse engineering of this.

Alex Zhang [00:27:54]: Yeah.

Swyx [00:27:54]: I’ve actually. I think calibration is probably the under. Like, people are just using it as a very fast classifier But they’re actually not even using the probability or calibration estimates.

Alex Zhang [00:28:04]: Yeah.

Vibhu [00:28:04]: I think it’s also still just misunderstood to reiterate. When you ask a model, “How confident are you?” it will spew out what, forty-three percent. the big delta is this is a grounded classification, right?

Alex Zhang [00:28:16]: Yeah. Yeah, I’m, I’m very excited to see what people do with this model. is it gonna solve everything? Like, no, of course not. But I think it solves a class of problems that we traditionally struggled with, which is low-latency things. So I love the examples with games. That’s actually like. I

Swyx [00:28:34]: Yeah, the Doom example

Alex Zhang [00:28:35]: Yeah

Swyx [00:28:35]: Was very good.

Alex Zhang [00:28:35]: I have a benchmark on language models playing video games. I’ve always been fascinated by whether you can have an intelligent system play new games and things of this nature. And so I think it’s really cool that they have sort of a unique way to do this, to capture language and understanding in, like, a fast, a very fast model.

Language Models Playing Video Games

Vibhu [00:28:59]: Oh, while you’re on the topic, anything you wanna point out for video games?

Alex Zhang [00:29:02]: Oh, yeah.

Vibhu [00:29:03]: This is. you did do a benchmark on any project, right?

Alex Zhang [00:29:05]: So, yeah. I guess these numbers are very outdated

Vibhu [00:29:08]: Yeah.

Alex Zhang [00:29:08]: Because a lot of the models are very different. And I’ve seen actually people run. There are some folks out there that are running newer models on these games, which is really cool. I guess the general premise of this benchmark was we just want to see if vision language models are good enough at just, like, plugging into games with the latency constraint included. ‘Cause this actually. I came out with this right after Claude Plays Pokémon came out.

Alex Zhang [00:29:33]: So this was, like, two years ago, which I guess is, like, ancient now. But I think what’s really cool about this suite of tasks, it’s very diverse in terms of what games they are. And also, I think most of the games are games that people know or, like, have seen before. I saw, yeah, Jeff playing Doom. I will say I don’t think. I think they were just playing, like, really simple levels and stuff. But honestly, like, most models still can’t really do. Or I don’t actually think any models can solve these games very meaningfully. Like, there are some games that they can. I think I’ve seen Astra be able to solve the Kirby game. And we also, for this benchmark, we intentionally designed a really minimal harness. And I will get to this point about harnesses because I think there’s a whole conversation to be had about, like, what is the value of a harness? Like, what is even the purpose of a harness for a model? But in general, like, I think it’s. yeah, I hope to see very quickly or very soon, like, all of these games beaten by newer models.

Vibhu [00:30:35]: Yeah, it’s interesting. Like, the old Cloud Place Pokemon, they, like, read state from RAM and saw what tiles are walkable and whatnot. We did a podcast with them A long time ago.

Alex Zhang [00:30:45]: Gotcha.

Vibhu [00:30:46]: Yeah. Just fun.

Swyx [00:30:46]: Yeah, and it’s similar. Like, Jeff doesn’t have vision

Alex Zhang [00:30:48]: Yes

Swyx [00:30:48]: So you have to kind of feed in,

Harnesses as Compositional Generalizers

Alex Zhang [00:30:50]: Yeah

Swyx [00:30:50]: These, like, game state and all these things. let’s go right into the harness stuff

Alex Zhang [00:30:54]: Awesome

Swyx [00:30:54]: Because you brought it up. Language model harnesses are compositional generalizers.

Alex Zhang [00:30:59]: Yes.

Vibhu [00:31:00]: You struggled to read that one.

Alex Zhang [00:31:01]: Explain. Yes. Okay. So I have been a little unsatisfied maybe with how people think about harnesses, because people compare like, “Oh, like, I love Claude Code, I love Codex, I love Pi.” Like, “No, I love Oh My Pi, I love Prime Agent.” To be honest, I think all of them are the same. Most of the design decisions or, like, the design choices around these harnesses are the same. Maybe Prime Agent is a little bit different because it’s, like, inherently an RLM. But in general, like, I think we can be a lot more creative with harnesses. And what by that is if we think about this from the perspective of what exactly is the harness doing for the model? Well, basically, when you’re trying to solve a problem and you want to use a language model to solve it, like a very difficult task, one thing that we have discovered is that next token prediction is a really awkward form to do a lot of these tasks. So for example, take SWE-bench. When you’re navigating a code base, like, are you going to be able to figure out how to do all of this with a single language model call? Like, you just say, “Solve code,” or like, “Solve my query over this code base.” No. So we rely on a harness to help you do these kinds of things. And I think what is interesting is, like, a harness is a very opinionated program over how you want a language model to be form fit over a problem. And I bring this up because when we think about the, what a harness is doing, we should really think about, like, what choices in the harness let the language model solve this task? And can I actually just have a language model that just does this? Because a harness, if you think about it, now that loop transformers are a thing, I think what’s really interesting about it is you can model a looped transformer in some ways, like, with a harness as well, right? You’re just looping over the model. Now you can say like, “Oh, I’m not decoding,” so it’s, like, a little bit different. But in general, we, for whatever reason, have stuck with the same model architecture choice forever. And I. And there’s many arguments for why, but clearly, like, we are now training models around harness rollouts. And so there is this very awkward way of doing training over harnesses, which is that we train a language model to act within a harness, but, like, now it’s like a really long, maybe, like, multiple agent rollout that we’re doing. And there’s, like, really awkward, hacky ways of doing this. So what this blog talks about is like, well, one way you can think about what is going on here is if the harness is basically helping the model solve a particular task, can different harness design choices actually do something a little bit more meaningful beyond just, “Here are some tool calls that will help you. Here is a way to grep through your code base.” And so this actually. The idea for this blog came with the RLM idea as well. We just didn’t package it that way. And I think this is actually true. There are many other ideas around RLMs that, like, we will be coming out with, but were all there from the beginning. these are all design decisions around. I think with what is. What I like about the RLM is that there were many iterations and versions of different abstractions that I was interested in doing, and ultimately the RLM made the most sense. But there’s a lot of reasons that aren’t public as to why that’s the case. you will see. So in this example, one of the things that we see with an RLM is that if you sufficiently offload context and write. ask the model to write code over that context, you get this really weird but useful property, which is that when the model recognizes during training how to solve a task, it turns out that the solution to many tasks is very similar across, like, tasks where you don’t even. Like, it’s not even that clear to you that the solutions are similar. So in this example, we have, like, a retrieval task and we have, like, an aggregation task, and they’re very different query. Like, the domain is just completely different. And when you train a regular language model over these two tasks, the trajectories look very different. And so what you’re relying when you. Like, let’s say you use Pi or Claude Code or something, which is not in this blog, but we do have these results. You’ll find that, like, these harnesses distinguish too much between these problems, even though the solutions are the same. And so one thing that we find when training RLMs is that, like. When it sees these problems, it’s the same. And the reason it sees these problems as the same is the sub-agent sees different problems, but the sub-agent is solving an easier sub-task, and so you’re confident it’s smart enough to do it. But for the base overall strategy, they end up looking the same. And so when you train on the left task, for example, it can immediately solve the right task. And so if you go down to, like, the plots that we have, one thing you’ll kind of observe is that the. as you just naively train your RLM on these tasks, they naturally learn to generalize, for example, to longer tasks, because the strategy is basically the same. You’re just modifying, like, a length variable. And this actually also holds for tasks that are different, and they’re, it’s not even-- they’re not different across length. They’re completely different tasks, math tasks versus writing tasks. But the solution, the, like, meta high-level solution is the same. And so when you train the RLM on one of them, it generalizes this behavior to the second one. And there’s no magic here. I guess maybe that’s the thing that I wanna kind of stress. Like

Vibhu [00:36:38]: How would you kind of verbalize what they are learning? So I think in here you say you train it on short tasks, they generalize to stuff 8–30x longer.

Alex Zhang [00:36:47]: Yeah.

Vibhu [00:36:48]: They are learning how to solve these type of problems, or what’s the, what’s the core thing they’re actually learning?

Alex Zhang [00:36:52]: Yeah. They’re learning how to solve these types of problems at a certain length. And it turns out that when you take the strategy that they learned, it is directly transferable to the longer length. Like, they’re

Vibhu [00:37:05]: Yeah.

Alex Zhang [00:37:05]: Effectively the same program. And this is something that is really exciting because what this kind of implies is that when you have a corpus of data or environments that you train your model on, the hope is that, like, A, you can train on less environments and generalize to more than what existing models can do through or harnesses through existing kind of, like, naive training. But B, also, you still wanna use all the data you have. So when you train on these tasks, like, hopefully it generalizes to a wider class of problems. And why this is also even more exciting, at least in the context of RLMs or any recursively calling system, is this argument holds inductively. So, like, I’ll give you an example, because I said, I claimed that competitive programming and GPU optimization use very similar skill sets. The model can. the harness potentially, you might have to nudge it a certain way, but it can learn that, like, “Okay, how I’m gonna go about solving this GPU programming task is very similar to what I learned for competitive programming. So I’m gonna list out a set of solutions. I’ll, I’ll, like, spawn subagents to list out promising solutions, and then I’ll, like, write this loop to go through and check these solutions, maybe evolve them, and, like, evolve them against a verifier.” And between these two tasks, this looks the same. But what the subagents are doing are maybe, like, unique and something, like, different. But even what the subagents are solving might actually also be of the same form, right? Because it’s like a, it’s a recursive argument. And so what I’m trying to get at with this whole blog post is just that, like, we should rethink what the role of the harness is, because harnesses can actually greatly increase the generalization capability of your model and the amount of data that it’s given. And this is not exclusive to RLMs. I think there is a wide class of harnesses that are yet to be discovered that actually can yield similar properties. And to extend this argument a little bit further, I think what you can also extrapolate from this is, like, if I look at an RLM, what are the components of an RLM that are actually necessary, and can I actually just directly train a model to do this? Can I train a model to act as an RLM implicitly in its forward pass? It’s a really weird thing to think about because, like, you might say like, “Oh, code is non-differentiable, blah.” But there are many approximations of this behavior that we will start to uncover. And I think, like, we will see beyond just, like, I’m gonna design a new coding harness that uses a special form of compaction or something. I think we can be a lot more creative here. Like, there’s, there’s so much we can do with these language models that I think we are just not doing. And I’m, like, very excited about this because I think, like, I think we can get serious gains from very opinionated and good harness design that lends itself better to scale. And what by this is, like, the RLM, for example, is a very primitive inductive bias. Like, there’s nothing super special about the design other than the fact that it’s very different than what we currently do. But this may potentially scale much better with, like, the data and the environments that we have available to us.

Vibhu [00:40:20]: I guess the, opposite thing that people would probably ask is current harnesses Are very generalized towards coding, which people see works for a lot of domains. Cloud code is being used for design, presentations

Alex Zhang [00:40:33]: Yep.

Vibhu [00:40:33]: Everything. MuseSpark, Grokbot.

Vibhu [00:40:36]: These are very simple, non-opinionated harnesses that are good at code, and that is also scaling out. what’s the example of how we improve those, I guess?

Alex Zhang [00:40:48]: Yeah. let me bring up another paper, which came out very recently. It’s like the harness tax paper. I think it’s by Arena. I really like this paper because it puts forward a prior that I had, which is basically that, like, most

Vibhu [00:41:07]: So it confirms the prior.

Alex Zhang [00:41:08]: Yeah. Like, most harness choices don’t matter because

Vibhu [00:41:12]: Yeah.

Alex Zhang [00:41:13]: All of these harnesses are the same. But I will say, like, Grokbot, for example, is actually quite different, I think, from my understanding, than how some of these other harnesses have been designed, and I like that a lot. and I think it’s clear from here at least that, like, I’m pretty sure. Anthropic or OpenAI are exclusively training on their harnesses. They’re probably not training on their competitor’s harness. I’d assume not, because I don’t know why they would do that. But

Vibhu [00:41:38]: But, this is a thing you see in open models, right? Like, Qwen is really good at using open code.

Alex Zhang [00:41:43]: Yes.

Vibhu [00:41:43]: They need to train in harnesses. Old Gemmas were notoriously bad at this.

Alex Zhang [00:41:47]: Yeah.

Vibhu [00:41:48]: Models are good, but you need to train in a harness.

Alex Zhang [00:41:50]: I think, though, as models get smarter, or, like, as they get better, this distinction becomes, not that important in the sense that, like, if you take Astra and you put it inside of open code, like, it’s not gonna go crazy, because I think it’s, like, just sufficiently good. And so I say this because the only benefits between these different harnesses is just cost, for the most part. And I think, like, what I’m getting at, Sugru, is if you plug these models into RLMs, though, they’re not that good still. They’re okay. And I think it’s mainly because the types, like the class of harness that we are training around is this class of harness, this, like, pi loop, this, like. I like to call it trajectory as a prompt, which just means, like, you keep the whole trajectory of the rollout as the context that your main model is using. Even if you use subagents, it’s still, like, kind of this form. And I think we’re going to. If we want to explore new harnesses, like, there needs to be teams that are dedicated to actually running meaningful experiments over, like, scaling out new harnesses, like maybe post-train scaling out on different harness designs. Like, I think we can actually get very meaningful knowledge or gains from doing this kind of thing, whether it’s an RLM or whether it’s something different. And that’s exciting, ‘cause I think, for example, if you train a lot on. Fable for a long time was the best model for RLMs because they had dynamic workflows, and it was pretty obvious that, like, this was a capability that was somewhat trained in. Even if the model was still, like, a little dumb, like, in the RLM harness, it still worked a lot better than other models did. Astra is now also, like, good enough at doing these things. But

Swyx [00:43:33]: Wait, is this where we see that Fable is the best for RLMs, or is there some other

Alex Zhang [00:43:38]: Oh, no, these are all internal results, I guess.

Alex Zhang [00:43:40]: Yeah, I don’t, I don’t have them

Swyx [00:43:41]: Okay

Alex Zhang [00:43:42]: Public right now. But in general, like, I think you can, You can very easily tell that we have not optimized for RLM, like, workflows yet. And I think if we get models that do this correctly, they will be a lot more efficient as well. you can kind of just think through, like, why this is the case, right?

Vibhu [00:44:03]: I think this is the point where you have to give the ten-second what are RLMs.

What Is an RLM?

Alex Zhang [00:44:07]: Oh, yes.

Vibhu [00:44:07]: Because there’s a lot of listeners here that

Swyx [00:44:09]: Yes. We’re, we’re assuming a lot of knowledge.

Alex Zhang [00:44:10]: Yes.

Swyx [00:44:11]: Also, I think you. But you have set some context

Vibhu [00:44:13]: Yes

Swyx [00:44:13]: So you can. Like, with everything we just said

Vibhu [00:44:15]: Yes

Swyx [00:44:15]: Can we have a clean, crisp definition of RLMs?

Alex Zhang [00:44:18]: Yes. Okay. I want to go back to the blog, the

Vibhu [00:44:21]: Yes

Alex Zhang [00:44:21]: The compositional generalizers blog. This one. Okay.

Swyx [00:44:24]: Okay.

Alex Zhang [00:44:24]: This is, like, the best. I wish I had this in the original paper. An RLM is basically just a harness design where the only tool in the harness is code, which is this programmatic subagent calling thing, where it has the option to call itself as a tool, and it has other tools. But all of these things are functions in code, and the context that it’s dealing with is always stored in some memory inside of this code environment. So this could be a file system. Like, this could be, like. I’ll give you an example, Prime Agent. The trajectory of Prime Agent, like the context, even the. when you compact and do all these things, is stored on disk. So the model can always reference its original context, even if it’s compacted, and all of its tools are run inside of, let’s say, like, a Python REPL or a Bash REPL. And so this. it’s like this very primitive abstraction. And I would say, like, where most harnesses differ is, A, context offloading is not done that, like that. if you look at Prime Agent, by the way, Prime Agent does context offloading, but not all the way. So, like, it still maintains the standard cod code, Codex loop of, like, trajectory as a prompt where you compact, but it has the additional kind of, like, the context is offloaded, and it only has. The unique point of Prime Agent is that the only tool is IPython. So this is, like, the very kind of generic abstraction around, like, RLMs. Yeah.

Vibhu [00:45:59]: Concretely, what’s the core thing RLMs are trying to solve?

Long Context, Composition, and Locally In-Distribution Tasks

Alex Zhang [00:46:02]: Yeah.

Vibhu [00:46:02]: At one point, I think when it first came out, it was context.

Vibhu [00:46:06]: I’ll pass the question.

Alex Zhang [00:46:06]: Yeah. So the original problem was harnesses have a really bad time dealing with long context. They typically were used to only really do them for, like, specific things like code. Like, they could deal with your code base because it was trained on it. But now it’s more around what this blog is talking about, which is Compositionality and the fact that, like, I think harnesses. We want to have language model systems that have much more control over the actions they make at every step. And what by this is tool calls are very limited because you have to invoke them every turn. Like, you have to invoke tool A, then tool B, then tool C, and there’s no central context that you can kind of draw back from. And RLMs are specifically, like, designed around composition and having, like, a central context that you can always draw from. and this context is, like, designed around the existing language models. Another, like, very similar example actually in design is, like, agent swarms, for example, the Hugging Face incident. Like, these agent swarms have, like, a message board that they learn to communicate over. And this message board, in some sense, is the shared context that they, like, act over. And RLMs basically say that, like, the best way to communicate through this is in code. Like, you write the code to do this, and it’s, it’s because these models are so good at writing code. like, we wanna take advantage of that fact.

Swyx [00:47:37]: And then for this compositional thing, up to and including generating your own harness, specific for the task.

Alex Zhang [00:47:44]: Exactly.

Swyx [00:47:45]: Right?

Alex Zhang [00:47:46]: I think we will start to see that if you go up to, this figure. Okay. So we talk about this idea of locally in-distribution tasks for a harness, and it is like a, an idea on top of, like, in-distribution tasks. When we think about language models, an in-distribution task is just a task where, like, the prompt is something that the model has either seen before or, like, has seen some version of it. Most harnesses work like the one on the right, which is they keep appending the trajectory as a prompt. And so eventually, unless you’re Anthropic or OpenAI and you train on, like, kind of these, like, user trajectories, most of these things end up being out of distribution for the most part. But locally in distribution is basically the compositional argument of if an RLM breaks down its computation into, like, kind of a meta-harness of sorts or, like, a program that involves subagents that, like, look at a local problem, every individual language model call over the course of this task is in distribution, even if the entire task itself is out of distribution. And this is a very desirable property, I think, for obvious reasons. Like, if every task is in distribution for each individual language model call, you will probably get to the right answer. so.

Swyx [00:49:00]: The logical limit of RLMs is RLLMs where, like, you not just, you don’t just write the harness, you also train a custom model for

Training RLMs and Smarter Harnesses

Alex Zhang [00:49:11]: Exactly.

Swyx [00:49:11]: You collect data, everything.

Alex Zhang [00:49:14]: Yeah.

Swyx [00:49:14]: Like, it’s a fully automated AI researcher inside of your harness.

Alex Zhang [00:49:16]: Yeah. We will see where the training of RLMs goes. I will say, as an academic, I am not working on this at MIT, or at least in the scaled sense, because I can’t afford to. but there are companies out there that are working on this. I think Prime and Select is very clearly working on this, and it’s very cool. Like, I’m, I’m very excited to see. Maybe we’ll observe, I don’t know, but maybe we’ll observe better, like, post-training scaling laws with when you train around a smart harness. Maybe we’ll even see smarter harnesses that come out and, like, they work better around these kinds of principles.

Swyx [00:49:50]: What is a smarter harness? Like, that doesn’t mean anything. You just said they’re all the same.

Alex Zhang [00:49:54]: No. What, more of what is, like, Claude Code, Codex, Pi, et cetera, are all the same in that, like, when you break down the logic of the harness, it’s, like, virtually the same thing.

Swyx [00:50:06]: Yeah, two calls in a loop or

Alex Zhang [00:50:07]: Yeah

Swyx [00:50:08]: Whatever.

Alex Zhang [00:50:08]: But With RLMs and with other harness abstractions, it looks very different. And this is where I think you really distinguish. it’s, it’s in the same way that, like, I think with language model architecture choices, a lot of architecture choices end up kind of looking the same when you, like, scale it out or, like, it. The differences end up being, like, somewhat minor in terms of. for a lab it’s not minor, but, maybe one model converges better than the other one, like, slightly. But in general, like, if you were. So for example, pre-training scaling laws only hold because the architecture choices we have are somewhat stable right now. But if you were to completely change the architecture, pre-training scaling laws probably don’t hold. Or, like, these kin…

Three months ago Dwarkesh, who has been posting incredible blogs and episodesabout RL, posted a framing question for his video essay on RLVR which upset a lot of Computer Use folks:

We are no strangers to learning in public and are no strangers to the stress of getting things wrong when you have a big platform. However, we were at Anthropic for the Computer Use launch, there for Claude Cowork with the first big podcast on it, organized the first Computer Use track at AIE presenting the state of the art, and were close to the OpenAI-Sky Software acquisition that now powers the complete domination of computer use that Codex enjoys today. This is why we’re excited to bring you today’s first guest, Ari Weinstein, cofounder of Sky and now leading all the amazing CUA progress that casuals might miss:

Ari explains why Computer Use is now “180 degrees different” from where it was months ago, how agents are learning to debug and recover from failures, why combining screenshots with accessibility data, the DOM, Playwright, and generated code changes the speed equation, and why the next frontier is making agents literally superhuman at using software.

OpenAI clones Jev

In the second half, Nikunj Handa from OpenAI’s API team breaks down the new developer stack: async tool calling, mid-turn steering, WebSockets, UltraFast inference, the Decisions API, prompt caching, pre-warming, compaction, and the Agents API. Given that we were the first Jev podcast, we particularly focus on the unusually fast sprint on the Decisions API:

And why it is just a Luna wrapper for now but the team is motivated and egoless enough to clone what they consider to be good patterns.

We discuss:

* Why OpenAI thinks Computer Use has changed dramatically in just the last few months

* Dots and what changes when every agent gets its own Linux computer

* Why Computer Use can now complete some tasks faster than the average human

* The path from human-level to “literally superhuman” computer use

* Why modern agents are much better at debugging and recovering from failure

* How screenshots, accessibility trees, the DOM, Playwright, and generated JavaScript work together

* App Shots and why they give models much richer context than ordinary screenshots

* Why Computer Use can close the loop between writing software and testing it

* Trust, permissions, and safety when agents can make payments and operate websites

* Async function calling and why models no longer need to stop reasoning while tools run

* Mid-turn steering, WebSockets, and the architecture behind more responsive agents

* UltraFast inference and how OpenAI is pushing frontier models toward much lower latency

* The rapid internal story behind the Decisions API

* Why Decisions API is more than structured outputs at low latency

* GPT Live, fast tool calling, and real-time computer control

* How OpenAI is already using Decisions API for support classification and internal workflows

* Longer prompt caching, cache pre-warming, and cache-aware applications

* Server-side compaction vs manual compaction for long-running agent threads

* What should live inside an Agents API versus a developer’s own harness

* OpenAI as an “AI cloud” and the search for higher-level primitives beyond raw model APIs

Ari Weinstein

* Product & Engineering, Computer Use at OpenAI

* X: https://x.com/AriX

* LinkedIn: https://www.linkedin.com/in/weinsteinari/

Nikunj Handa

* Product, API at OpenAI

* X: https://x.com/nikunjhanda

* LinkedIn: https://www.linkedin.com/in/nikunjhanda/

Timestamps

00:00:00 OpenAI DevDay: Dots, GPT-6.1, Agents API, and Decisions API

00:02:52 Dots and Personal Cloud Computers

00:04:59 Why Computer Use Is “180 Degrees Different”

00:06:04 From Sky to Self-Debugging Computer Use Agents

00:09:24 How Computer Use Sees and Operates Software

00:12:09 From Faster Than Humans to Superhuman Computer Use

00:16:03 Agents API: Trust, Permissions, and Safety

00:17:31 Computer Use for Coding, Testing, and QA

00:19:14 GPT-6 APIs, Async Tool Calling, and UltraFast Inference

00:23:21 The Rapid Story Behind Decisions API

00:25:32 What Decisions API Is and How It Works

00:30:24 What OpenAI Is Building With the New APIs

00:32:23 Prompt Caching, Pre-Warming, and API Performance

00:35:20 Context Compaction for Long-Running Agents

00:37:13 Memory, Higher-Level APIs, and the AI Cloud

Transcript

Introduction: OpenAI DevDay and the New Agent Stack

Vibhu [00:00:00]: Okay. We’re very excited to be here. Today is OpenAI DevDay. Special podcast

Swyx [00:00:08]: We’re the first podcast after your livestream.

Vibhu [00:00:10]: First podcast. We have Ari here, who leads the product and engineering team for Computer Use agents. Before we kick in and dive deep on Computer Use, you wanna give a quick recap? What was announced? What’s the quick slew of announcements you guys had today?

Ari Weinstein [00:00:24]: Yeah. yeah, it was a super exciting day. we just got out of the keynote. It was really sick. there were a bunch of Computer Use announcements that I think are worth thinking about. We have, Dots, which is the new, sort of personal assistant product, and, that has some really exciting Computer Use features. There’s GPT-6.1 Sol, which is this amazing new model, that I think is particularly great for Computer Use ‘cause of, sort of the cost and speed, advantages. I think, I think we shared that it’s, a fifth of the cost of Astra and a seventh of the cost if you’re looking at Computer Use specifically, which is really amazing. sorry, there were so many things. I’m trying to sort through it.

Swyx [00:01:02]: And the API.

Ari Weinstein [00:01:03]: Agents API, which now has Computer Use in it, which is really cool, ‘cause now developers can build on the same Computer Use, that is part of Codex, and ChatGPT. and then there were some demos of our existing Computer Use features, like app shots, where you can take the context of something you’re doing on your computer and bring it into Codex and ChatGPT really fast. And then, like, native Computer Use on your Mac, where Roman had it taking screenshots of his app, automatically, and he could do other things on his computer while Computer Use was using his applications. so yeah, really exciting keynote.

Swyx [00:01:35]: And not to mention the Decisions API.

Ari Weinstein [00:01:37]: Decisions API.

Swyx [00:01:38]: Off the bat, are they all the same model? Like, this is. Or the same dataset distilled to different models?

Swyx [00:01:44]: Like, basically, like, is Computer Use using Decisions API, or are they, like, kinda separate?

Ari Weinstein [00:01:49]: So what’s really cool about the Decisions API is it, you know, it has all these new capabilities. It does inference in parallel. it doesn’t have reasoning. It’s a smaller model, than the ones we use for Computer Use. and so those capabilities make it really fast.

Dots and Delegating Work to a Cloud Computer

Swyx [00:02:07]: Yeah.

Ari Weinstein [00:02:07]: They also make it a little bit less good at doing, like, long horizon, sort of sophisticated tasks. And so I think I would say it’s still an open area of research for how we, like, bring those approaches together. But, yeah, I’m really excited to see what people build with the Decisions API.

Vibhu [00:02:24]: One of the interesting things is Dots now have attached personal computers.

Ari Weinstein [00:02:28]: Yeah.

Vibhu [00:02:28]: So it seems like they’re very much more persistent. You’ve been using them for a while. How should people push the bounds? Like, what should people aim for? What should they try? Personally, right now I use it for a lot of customer service. Like

Ari Weinstein [00:02:41]: Cool

Vibhu [00:02:41]: “Oh, this was wrong. I don’t wanna sign in. I don’t wanna authenticate.” Find whatever and just get it fixed.

Ari Weinstein [00:02:45]: Yeah.

Vibhu [00:02:46]: How should we push further? What should people try?

Ari Weinstein [00:02:50]: Dots Are a really cool product because each Dot has access to its own Linux virtual computer in the cloud, which is different from our other products. you know, traditionally, we’ve have access to a browser in the cloud, or it has access to your own computer, but now you get your own entire Linux computer in the cloud. And so it can run full desktop applications, and it can also use a web browser. And so, yeah, you know, I think the powerful thing about Computer Use and the reason why I think it’s so, exciting is because it makes it so that the agent can do anything you as a, as a person can do, because all the software in the world was designed for humans, and now agents can use that same software, and you can delegate to the agent. So, yeah, like, anything that you would do on a computer, you can ask a Dot to do. Yeah, I think what particularly is useful is gonna really depend on who the end user is and what- what’s valuable in their life. but yeah, I would just start by thinking about, like, one of the things that you spend time on and how could you delegate those to an agent.

Swyx [00:03:47]: Yeah, a lot of flight booking and shopping and honestly even, like, playing a game or whatever, right?

Ari Weinstein [00:03:52]: Totally.

Swyx [00:03:52]: Yeah.

Ari Weinstein [00:03:53]: Yeah, I don’t know. For me, something I did recently, I’ve been working on. I’ve, subscribed to a meal prep service ‘cause I was trying to, like, eat healthy, you know? And I really like this meal prep service I found because it lets me customize the meals I order to, like, a high degree of granularity. So I can say, like, “I want this many grams of chicken and this many grams of rice.” but it was so complicated. It took me two hours to do an order, and I found that I could ask Computer Use to do it for me, and it did it in 15 minutes. so I actually saved two hours. it both did it eight times faster than I could, and it saved me two hours on GPT-6.1 Sol.

Swyx [00:04:32]: Yeah.

Ari Weinstein [00:04:32]: So those are the kinds of tasks that I feel like, are really powerful.

Swyx [00:04:36]: As a creator, I can tell you automatically, immediately, my number one use case is automating YouTube.

Ari Weinstein [00:04:40]: Nice.

Swyx [00:04:40]: Because, YouTube doesn’t expose a lot of things via API.

Ari Weinstein [00:04:43]: Yeah.

Swyx [00:04:43]: And you have to just put it in a VM and just, like, run it, for, like, let’s say, let’s say their AB testing feature or making community posts. None of this is available by API ‘cause they hate developers.

Swyx [00:04:53]: Anyway, so,

Ari Weinstein [00:04:55]: I’ve heard that from our developer experience team too. They use it with YouTube a lot. Yeah. It’s really awesome.

Swyx [00:04:59]: So I wanna draw for, you know. let’s say, I wanna get a little bit spicy. One of our, the leading AI podcasts, our friends, is famous for saying that Computer Use hasn’t advanced in the last two years.

How Computer Use Has Changed in the Last Year

Ari Weinstein [00:05:12]: Yeah.

Swyx [00:05:13]: Which is a very interesting statement, and I think you’re one of the best people in the world to talk about this, like, how have things have progressed, right?

Ari Weinstein [00:05:20]: Yeah. You know, they said that a few months ago, I think, and I hope they have a different perspective now because Computer Use is, like, 180 degrees different than it was.

Swyx [00:05:26]: He’s a, he’s a tough guy to impress.

Ari Weinstein [00:05:27]: Yeah, okay. well, we’re working on it.

Swyx [00:05:30]: But, you know, you worked on. You’ve, like, basically spent your whole career working on, like, some kind of computer automation, right?

Ari Weinstein [00:05:34]: Yeah.

Swyx [00:05:34]: Like shortcuts

Ari Weinstein [00:05:35]: Yeah

Swyx [00:05:35]: At Apple, and then Sky, and then, and then joining OpenAI. Can you draw, like, what your through line is for, like, what is driving you and what- you, what wasn’t possible back then maybe

Ari Weinstein [00:05:47]: Yeah.

Swyx [00:05:48]: And, like, what your sort of milestones were.

Vibhu [00:05:49]: I guess to add on to that as a follow-up question, what’s the major change from using Codex Computer Use from, like, last week

Ari Weinstein [00:05:57]: Yeah

Vibhu [00:05:57]: Through to today? Is it model? Is it dots? Is it harness? So all the history plus what really just changed in today’s announcements?

Ari Weinstein [00:06:04]: Yeah. On the through line, I guess I’ve always been excited about automation and helping people automate tasks because then you can, like, save time in your life and focus on things that are more important to you than, like, operating a computer very intricately. And so, yeah, that was why we worked on some of those products. I was at Apple before. we made a company called Sky. we ended up joining OpenAI, which is really exciting. and I think something that was

Swyx [00:06:27]: And almost like you have to hack around Apple until Apple was like, “Fine, like, we’ll just hire you and you can just work on the inside,” right? Like.

Ari Weinstein [00:06:35]: It was, it was a cool place to get to work. what was really interesting looking back at Sky is we were, we were working on Computer Use there as well, and the models were so much less capable. And now the models, just in the last one year, have become extraordinarily capable at Computer Use. I think the biggest delta that I see is before they could, like, reliably start tasks, but then they would run into problems, and now they’re really good at debugging. They’re really good at trying again, introspecting what is and isn’t working. and I think we’ve also brought the Computer Use the Computer Use field itself has moved forward. I think we’re using more techniques. now Computer Use, often writes code. So if you actually look at it in Codex and you expand the tool calls manually, you can see that it’s not just doing one action at a time. It’s actually writing JavaScript code that it executes, that the computer executes to perform sometimes many actions at once, which is a great, you know, speed up and great capability. We use more accessibility, sort of multimodal interfaces. So, the model may use screenshots, it may use accessibility, it may use Playwright. it can use a lot of different mechanisms, based on the task at hand. and then, yeah, the model acceleration has been, has been just amazing. So, yeah, what’s different today? I think we’re making computers better all the time, so I think just, like, one day’s difference, is probably a little bit less consequential than, like, even the past month or the past two months. but, yeah, I think the Computer Use in Dot is really exciting as well as, the new model that we came out with.

Measuring Computer Use and Improving the Harness

Vibhu [00:08:03]: On the keynote, Tejal was mentioning 7x improvements in Computer Use speed, a lot better on a few benchmarks. How do you guys think about measuring it? Computer Use is one of those things where, as you say, you know, it’s improvements over time.

Ari Weinstein [00:08:20]: Yeah.

Vibhu [00:08:20]: Is it harness? Is it model? Is it post-training?

Ari Weinstein [00:08:22]: Right.

Vibhu [00:08:22]: How do you guys look at it internally about measuring how good it is, and what were the changes with the new model?

Ari Weinstein [00:08:29]: We actually have a bunch of different ways of measuring it, some of which are on different permutations and configurations of the harness. It’s a bit of a complicated story because, you know, our production products have, you know, some more safety checks, and, you know, those are configured differently based on the needs of the, of the task at hand. So there’s a lot of ways to measure it, but I think regardless of how we measure it, we find pretty consistent gains. and those gains are, sometimes in the harness and sometimes in the model. and yeah, I was really excited by this result that GPT-6.1 is even more cost-effective for Computer Use than its baseline cost improvement as compared to Astra. It’s, like, really cool to see.

Swyx [00:09:10]: Yeah. I mean, one of the visuals I really liked from the livestream was that, you’re sort of improving the Pareto frontier of, your, curve, and there was a lot of talking about how you’re improving it together with the harness.

Ari Weinstein [00:09:24]: Yeah.

Swyx [00:09:24]: Can you give some examples of aha moments that you had, whether it’s on, like, model driving the harness driving the model, whatever?

Ari Weinstein [00:09:32]: I don’t mean to repeat myself, but I think, like, introducing more modalities has been really powerful.

Swyx [00:09:36]: Okay.

Ari Weinstein [00:09:36]: One more specific example of that is, in the past, I think we saw a lot of Computer Use, products had to spend a lot of time, like, scrolling, you know? So it would, like, take a screenshot. It would try to do something. It would be like, “Oh, I gotta, like, scroll down to the next page of results,” and then it would take a screenshot, and then it would try to do something. It would scroll down again. And so I think, with accessibility and other. and, direct access to the DOM and other things like that, now the language model can actually see, like, an entire page or an entire application. It can write code that can do multiple steps at once. And so I think those have been probably the biggest single aha moments. There’s, like, a lot of tiny ones that are less exciting in comparison, but actually we do find also that a lot of speed improvements are driven by, like, a lot of little paper cuts that we gotta go in and introspect.

App Shots, Accessibility, and Better Computer Context

Swyx [00:10:21]: Yeah. A lot of really hard engineering.

Ari Weinstein [00:10:23]: Yeah.

Swyx [00:10:23]: I mean, app shots in general, right? Like, I think people don’t quite get the difference if. because there’s, like, a nice visual in Codex when it

Ari Weinstein [00:10:30]: Yeah

Swyx [00:10:30]: When you take an app shot, but they don’t maybe they get the difference that, you are able to actually drive each button and you have the, you have each text, in a very optimal representation.

Ari Weinstein [00:10:40]: Yeah. Exactly. Yeah. It’s kind of fun actually. If you wanna be, like, really nerdy about it, you can go into Codex, take an app shot by hitting the two command keys. So you grab the content from whatever app you’re working with, bring it into the, Codex or ChatGPT chat. And then the. if you click on the attachment and you click on this, like, little tiny button in the top right, you can see the raw text and you see the raw accessibility representation. And yeah, we’ve put a lot of work into, putting

Swyx [00:11:04]: Just dumping everything out. Yeah.

Ari Weinstein [00:11:05]: Dumping it out, but also making it token-efficient, doing it efficiently. There’s, like, a bit of an art to it. And, you know, it turns out that the same technology that was invented for humans, you know, who maybe have accessibility needs, who wanna use a screen reader technology, that technology is really helpful for them to be able to use computers. It’s also really helpful for LLMs to be able to use computers. So that’s been, like, really fun to get to work on.

Vibhu [00:11:27]: For context, I feel like a lot of people don’t understand app shots. They don’t even know it’s a feature.

Ari Weinstein [00:11:30]: Yeah.

Vibhu [00:11:31]: It’s when you double hit command, it pulls in what looks like a screenshot

Ari Weinstein [00:11:34]: Right

Vibhu [00:11:34]: And you’re like, “Oh, why have I opened up just a screenshot and thrown it in?” No, it’s actually pulling all the metadata, all the code, everything.

Ari Weinstein [00:11:40]: Yeah, exactly. Yeah. So it’s like, you know, if you take a screenshot of a webpage that has a link- The screenshot doesn’t include where the link goes. It doesn’t include, you know, maybe you take a screenshot of your calendar, the ca- event ti- titles are truncated, you know? But when you take an app shot, it gives, like, the language model, like, full context about everything and, that lets it, just sort of, like, do much more.

Swyx [00:12:02]: Yeah. For those who wanna see more, Jason Liu, I invited him to do a full workshop on this, at AI Engineer.

Ari Weinstein [00:12:07]: Amazing.

Swyx [00:12:08]: Did a great job.

Vibhu [00:12:09]: I have a broader vision question

Toward Superhuman Computer Use

Ari Weinstein [00:12:11]: Yeah

Vibhu [00:12:11]: On Computer Use agents. So your example of take a screenshot, scroll page, take a screenshot is where we were.

Ari Weinstein [00:12:17]: Right.

Vibhu [00:12:17]: Today, they can automate a lot. what are the bottlenecks? Is it models? Is it harnesses? What. Where do you see it going in, like, two years? Do you see it just running for hours? How do we get there? Any predictions on where Computer Use goes?

Ari Weinstein [00:12:32]: Yeah. I mean, I think what’s really crazy that I think, You know, the team’s accomplished over the past couple of months is that now Computer Use is, like, faster at accomplishing tasks than, like, the average human probably in most cases. and I think that the next frontier is to have Computer Use be, like, literally superhuman in its performance where it actually is as fast or faster at using software than, like, expert Computer Users like us. and I think that’ll be really consequential and exciting when that happens because I think we’ll be able to all of a sudden build products, that, provide just much more real-time experiences. And I think it’ll also. lowering the barrier to entry of, or the activation energy, I suppose, of using Computer Use I think will make us start to default to doing certain things in agents that we’ve become accustomed to doing manually. And I think that’s exciting also ‘cause it’ll save us a ton of time. and I think there’s a, you know, there are a lot of different little paper cuts and bottlenecks that are sort of standing in the way of that. I think that there’s, yeah, there’s things on the model side, there’s things on the inference side, there’s things on the harness side, there’s things in the, in the representation. You know, we find that as Computer Use gets faster, we’re increasingly bottlenecked by just, like, the speed of doing an operation. Like, for example, you know, a non-trivial amount of time in our benchmarks of Computer Use tasks is actually, like, let’s say you’re automating a task on doordash.com. Like, a lot of the time is actually waiting for doordash.com itself to load, you know?

Swyx [00:14:04]: Yeah, then you just write a wait and then you execute the wait.

Ari Weinstein [00:14:07]: Yeah, totally. And you wanna get. Yeah, actually, it’s actually really important that you get that de- like, you want as little delay as possible between when it finally finishes loading and when you go and

Swyx [00:14:16]: Yeah

Ari Weinstein [00:14:16]: Trigger the LLM to do the next action, which is actually- itself a statistical science.

Swyx [00:14:20]: Like an event-driven way maybe to do that.

Ari Weinstein [00:14:22]: When possible, you want it to be event-driven.

Swyx [00:14:24]: JavaScript has some load events.

Ari Weinstein [00:14:25]: And JavaScript has load events for. or the web browser has load events for web navigation, but there’s other types of events that actually really can’t be event-driven. So there’s a lot of complexity

Vibhu [00:14:34]: The one that comes to mind is, like, chatting with customer service.

Ari Weinstein [00:14:37]: Yeah.

Vibhu [00:14:37]: Replies could take 30 seconds, could take three minutes.

Ari Weinstein [00:14:39]: Oh, right.

Swyx [00:14:41]: I have dealt with so many bots with Codex. it’s great, but I also wonder if the other side knows that they’re talking to a bot ‘cause I’m, like, answering in complete sentences. Like, I’m capitalized correctly.

Ari Weinstein [00:14:50]: That’s hilarious.

Swyx [00:14:51]: Like, I’m giving full num- full reference numbers and everything. Like, it’s too. it’s clearly too good. I don’t care. Like Like, I’m just, like, trying to get my support case.

Vibhu [00:14:58]: I’ve prompted it to, like, you know, “Don’t pretend you’re a bot. Be very annoyed human.”

Vibhu [00:15:02]: Short one-liners, like

Swyx [00:15:04]: Yeah

Vibhu [00:15:04]: Push it, do all this. I also tell it, “While you’re waiting for responses, like, use subagents to research better ways to figure out what we need.”

Ari Weinstein [00:15:12]: Nice.

Vibhu [00:15:12]: It’s just, like, human little intervention.

Ari Weinstein [00:15:14]: That’s awesome. I also feel like half the time it’s a bot on the other end, so now you

Vibhu [00:15:17]: Yeah

Ari Weinstein [00:15:17]: Got the bots talking to each other.

Swyx [00:15:18]: Yeah. I will also say, you know, like, you know, one milestone of Computer Use that we are, we’re at now is, you know, three, four years ago, we were scared of hooking up LLMs to the, to the web and to

Ari Weinstein [00:15:31]: Yeah

Swyx [00:15:31]: To our, to our devices. And now I’m having it configure DNS for me.

Ari Weinstein [00:15:35]: Wow.

Swyx [00:15:36]: I’m having it pay my bills, and, like, really, like, tens of thousands of dollars of, like, stuff I’m just sending it over and yoloing with Computer Use and, like, you know, what’s the, what’s the worst thing that can happen?

Swyx [00:15:48]: So that- that’s all, that’s all really good.

Building Safely With Computer Use in the Agents API

Ari Weinstein [00:15:50]: Yeah.

Swyx [00:15:50]: I think now that you’ve. you know, obviously, you also have to dogfood your own products and all these things. Now that you’ve sort of released this in API, what are some pitfalls or tips that you wanna tell developers, because they’re about to, I guess, encounter all this, firsthand?

Ari Weinstein [00:16:03]: First of all, I’m just really excited that we brought Computer Use into the Agents API. I think this is, really great because obviously a lot of developers are building applications that wanna be able to work with third-party websites and services. And so Computer Use has this universality to it. It can work with anything. So now all of a sudden, developers can build using the same Computer Use implementation that we’re building on. I think there’s great work to be done if you wanna build your own Computer Use harness, but it’s hard. And also, we train our models on our Computer Use harness, so there is, an advantage to using the one that’s in distribution for the model. There actually might be a speed and cost and accuracy advantage. So I think it’s really great for people to get to build on top of that. And, yeah, you know, I think kind of to the point that you were making, like, I think we’re all sort of still in the process and maybe, like, some of us are ahead of many people in the world of, like, getting comfortable with this technology and trusting it. And so I think it’s incumbent on us to, sort of build that trust over time by making sure we’re building things that are reliable, by building, the right kinds of safety checks, by asking for the user’s consent before doing something consequential like making a payment, by, asking, you know, maybe depending on the application, making sure you’re only letting it access the websites or applications that it actually needs for the task. So that’s, I think, something important to think about. but yeah, I’d really encourage people to try the new Agents API, build all kinds of cool stuff on it. We’d love to hear your feed- feedback if, you know, depending on how it goes.

Vibhu [00:17:31]: Have you seen any changes in the way it affects dev workflows? So one of the things with dots is, you know, you’re seeing it in Slack.

Ari Weinstein [00:17:38]: Yeah.

Vibhu [00:17:38]: You’re seeing people use voice and build. the example Roman showed of change this app and send me screenshots along the way and all this.

Computer Use for Testing and Closing the Software Loop

Ari Weinstein [00:17:46]: Yeah.

Vibhu [00:17:46]: Is anything that you’re seeing there in adoption about how people are using Computer Use for coding workflows? Any tips people should take from that?

Ari Weinstein [00:17:55]: One of my favorite use cases for Computer Use actually, and one that we see a lot in the wild, is Computer Use letting the agent- actually test the software that the agent has built, which is far more consequential than it sounds. Because traditionally, you know, you’d build something in Codex and then the-- and the Codex builds it for you, and then you have to test it, and you are now like QA for the agent, right? So with Computer Use, you can complete the develop-- the software development life cycle, where, the agent can build software, it can test it. So I have a lot of fun, you know, building stuff, having the agent test it. By the time it comes to me, it’s already working. I have, extra fun because sometimes I’m, like, developing Computer Use itself, and so now I have a Computer Use agent that’s using my Computer Use agent that’s using something else. so yeah, I really, I really think this is a super powerful class of use case.

Swyx [00:18:44]: I have a visual play test skill that I’ve developed that, really catches a lot of design issues,

Ari Weinstein [00:18:49]: Nice

Swyx [00:18:50]: That, you know, normally when you just look at code, you wouldn’t really pick it up. it’s also really good for cloning apps, though. If you’re using a shitty SaaS and you wanna kill the SaaS You just clone it screen by screen by screen. and Obviously, Computer Use can completely drive everything, take screenshots, note it down, and then clone everything with Codex.

Ari Weinstein [00:19:06]: That’s really cool.

Swyx [00:19:06]: But yeah, thanks for all your progress. I think, that is

Ari Weinstein [00:19:08]: Absolutely

Swyx [00:19:09]: Our time.

Nikunj Handa: What’s New in the OpenAI API

Ari Weinstein [00:19:10]: Yeah.

Swyx [00:19:10]: This is not the last that we’re gonna talk.

Ari Weinstein [00:19:12]: Yeah, cool. This has been really fun. Thank you guys for having me.

Swyx [00:19:14]: All right.

Vibhu [00:19:14]: All right. Okay, we’re a strict cutoff. We’re just gonna dive right in.

Nikunj Handa [00:19:17]: Let’s do it, yeah.

Vibhu [00:19:19]: Okay, so, Nikunj, we’re very excited to have you. You shipped a lot on the API side, like we just

Nikunj Handa [00:19:25]: Yeah

Vibhu [00:19:25]: Talked about with Ari. You can now build with Computer Use agents. Anything you wanna highlight, the API side of changes, and introduce yourself a little and what you do?

Nikunj Handa [00:19:34]: Yeah, for sure. My name is Nikunj. I lead product for the API team. Been here for roughly three years. been working on launching models. I feel like that’s just been, like, a thing, constant thing throughout my time, here at OpenAI. And, with every new model, we try to, like, basically work super closely with the post-training team, the research team, to figure out what’s new in it. and then we, like, expose those capabilities in the API. so that’s, like, the basic way of putting it. and if you just look at, everything that’s new with GPT-6, the cool new capabilities that we launched were, firstly, async function calling. so what you see with, like a lot of the things that you’re seeing in, like, Codex and Dots and everything is that tool calls take so long that you don’t have to, like, pause the model’s execution while, the tool is running. So you could just, like, kick off a tool call, keep running, keep reasoning, and then check back in. so we launched async tool calling. We launched, like, mid-turn steering, so now you can, like, inject messages while the model is reasoning, in the middle. so as your tool call finishes, you can put in that instructions.

Async Tool Calls, Mid-Turn Steering, and WebSockets

Swyx [00:20:43]: And that’s also partially a model alignment capability, right?

Nikunj Handa [00:20:46]: Yeah.

Swyx [00:20:46]: Like, they have to train in the ability to train.

Nikunj Handa [00:20:48]: Exactly, yeah. And

Vibhu [00:20:49]: I feel like we’ve had it in the app. You could always, as it’s reasoning, you could steer.

Nikunj Handa [00:20:54]: Yes.

Vibhu [00:20:54]: It wasn’t the best. It’s gotten much better.

Nikunj Handa [00:20:57]: Yeah.

Vibhu [00:20:57]: Excited to see how it does this in version

Nikunj Handa [00:20:58]: Yeah, and I like our main

Vibhu [00:20:59]: And now

Nikunj Handa [00:21:00]: Goal in, our main goal in the API is to, like, put things in the API once it’s trained into the harness. And so we kinda wait for that moment until it’s good enough. And a lot of that is, like, actually being powered by WebSockets, which we launched, a few, I wanna say months ago. And so WebSockets just opens this, like, whole bidirectional, like, communication thing with the model. This is not, the GPT Life thing. I’m just talking about GPT-6. and you can do all these, like, async tool calling, async reasoning, injecting messages. It’s a really fun API to work on. I think, like, really enjoying.

Swyx [00:21:33]: Yeah. This is why we are the engineering podcast, because we get to talk about WebSockets.

UltraFast and the Inference Stack

Nikunj Handa [00:21:36]: Yeah.

Swyx [00:21:37]: This also pairs very well with UltraFast, right?

Nikunj Handa [00:21:39]: Oh, yeah.

Swyx [00:21:39]: Like, that is now, like, I think for the first time ever available in the API.

Nikunj Handa [00:21:43]: Yes.

Swyx [00:21:43]: Which is, which is basically the theoretical fastest speed you can ever get, Frontier of Intelligence.

Nikunj Handa [00:21:49]: Yeah. It’s been so exciting to work on that project. I think, before I go into the API, the most fun part of, UltraFast has been just watching the inference team cook with Astra. Like, they’re just, like, constantly having these, like, Codex agents running, trying to, like, squeeze out more performance. And, I would say, like, at least for a couple of months, a lot of it was focused on efficiency and driving the cost down, which is how we, like, were able to cut the Luna price by, like, 80%. It was, like, a lot of that was driven by, like, all the inference improvements they landed. And then now they’ve, like, shifted gears towards, like, how can we make this run as fast as possible? And so UltraFast has just been, like, amazing to see on a mo- on a model like Astra. Like, to go that fast has been really cool. And yeah, WebSockets is like. actually it was like the first time we launched WebSockets, it was for GPT, 5.3 Codex Spark, which was. Can’t believe we named a model that, but, you know, that’s what we launched it for. And obviously, it helps so much because, like, you gotta have the tool calls. you had, like, really reduced the overhead, of going back and forth with tools. And so, WebSockets is awesome for that.

Swyx [00:22:57]: Yeah. it’s always cute to see, like, I have my reset usage limit, and then I have my Spark usage limit that I never use.

Nikunj Handa [00:23:03]: Yeah.

Swyx [00:23:04]: Like, it’s there if I want it.

Nikunj Handa [00:23:05]: I think it’s gone finally.

Swyx [00:23:06]: It’s gone. It’s gone, yeah.

Nikunj Handa [00:23:07]: I know it’s gone, so.

Swyx [00:23:08]: Yeah. you’re slowly killing off all the, you know, the

Nikunj Handa [00:23:11]: The old ones, yeah.

Swyx [00:23:11]: Oldies.

Vibhu [00:23:11]: This is a great week. I mean, it was the first time we had Frontier Intelligence at extreme speeds.

Nikunj Handa [00:23:17]: Yeah.

Vibhu [00:23:18]: People really liked it.

Nikunj Handa [00:23:19]: Yeah.

Vibhu [00:23:19]: So

Swyx [00:23:20]: Yeah

Vibhu [00:23:20]: First time it comes back.

Swyx [00:23:21]: Yeah. for, 5.3 Spark is explicitly attributed to Cerebras. You guys are not confirming or denying that, UltraFast is related to Ce- Cerebras, but people are. I’ll just say that people do care and, are wondering about it. And you have your own silicon as well. elephant in the room, decision models.

Decisions API: OpenAI’s Fast Decision Model

Nikunj Handa [00:23:38]: Oh, yeah.

Swyx [00:23:38]: Decisions API. We were the first podcast to do a big Jev, deep dive with, Diogo, and I also, you know, featured him at AI Engineer. How quickly did you see Jev and go like

Nikunj Handa [00:23:49]: Oh my gosh. Yeah.

Nikunj Handa [00:23:50]: Yeah. Firstly, like, huge props to Diogo and, like, the Jev team for, like, really inspiring the

Swyx [00:23:55]: Yes

Nikunj Handa [00:23:55]: Like, whole segment in the market. Like, obviously Jev comes out, everyone’s, like, losing their minds over it. Our users are, like, hitting us up. But also, like, our internal teams are like, “We need, like, a much faster classification system.” We can. I don’t wanna, like, get ahead of some of the dots features that are gonna come

Swyx [00:24:16]: Whoo

Nikunj Handa [00:24:16]: But you’re gonna see, like, some cool, like, really snappy, fast things built on top of the decisions API. but, you know, like, yeah. Props to Jev for, like, inspiring this whole thing. obviously a bunch of people at OpenAI get nerd sniped by that, and they’re like, “How can we, like, make this work? We’re not gonna, like-”

Swyx [00:24:33]: Okay.

Nikunj Handa [00:24:33]: “. train a new model.” But

Swyx [00:24:34]: Like, four weeks ago, this was not on the dev radar, right?

Nikunj Handa [00:24:37]: No, not at all. No.

Swyx [00:24:37]: Okay.

Nikunj Handa [00:24:37]: This is like

Swyx [00:24:38]: Wow

Nikunj Handa [00:24:38]: Jev-inspired and, like

Swyx [00:24:40]: I think you are officially the first one to your lab to, like, clone and, adopt this.

Nikunj Handa [00:24:44]: Yeah. Yeah. I feel like, OpenAI has such a strong, like, hacker culture and, like, people are just, like, they get excited about things. And so, guy from inference, this one awesome guy from, the infra team are like, “ this is amazing. We’re gonna, like, hack on it.” They build a prototype, it, like, works, and now we- we are just, like, hill climbing on latency and trying to make this as fast as possible, and we wanna, like, launch it in the coming days. so as soon as we hit our, like, latency target, we’ll try to get this out.

Vibhu [00:25:13]: It’s interesting. At the same time of hacker culture, you also, as Sam said, like 99%, one of the most reliable APIs with

Nikunj Handa [00:25:20]: Mm-hmm

Vibhu [00:25:20]: I think probably the most usage, which is your team directly. how should people see decisions API? I feel like a lot of people saw Jev, heard the buzz, haven’t built with it. You’re making it very mainstream.

What Decision Models Are Good For

Nikunj Handa [00:25:32]: Mm-hmm.

Vibhu [00:25:33]: What should people see it as? How should they use it?

Nikunj Handa [00:25:36]: Yeah. I think the main use cases we’ve seen is, like, really fast classification. all the Computer Use demos have been amazing and really cool. I think there will be limitations, of course, in terms of, you know, having Astra, like, write, like, a JavaScript-like script to control your computer, versus having Luna pick, like, one action at a time. I think, it’s not gonna be at the same intelligence level, but, like, maybe there’s some Computer Use tasks that this is good enough for. So excited to see that come through. the other cool prototype I’ve seen internally is people hooking it up with GPT Live. So GPT Live is like, you know, our bidirectional, like, real-time,

Swyx [00:26:14]: Voicing

Nikunj Handa [00:26:14]: A- API. And, it’s built on this, like, model of front-end models and back-end models. So GPT Live is this, like

Swyx [00:26:20]: Think or talker

Nikunj Handa [00:26:21]: Super fast. Yeah, think or, talker thing. So GPT Live is the talker, super fast, really good at delegation, and you have something like Astra sitting at the ba- at the back. But tool calling has always felt, like, really slow in GPT Live. and so people have been, like, putting together these, like, tool calling demos of GPT Live controlling a computer, and it just feels like so much more snappy and natural. So I’m, like, kinda excited to see, like, what people do with Live and with Luna on decisions API. so that’ll be pretty exciting. Yeah.

Swyx [00:26:55]: So I wanna iron this out for people, especially from the product side, because a lot of people have been putting out Jev clones. There’s been about 100 in the last two weeks.

What Makes a Decision Model Different

Nikunj Handa [00:27:01]: Oh, really? That’s amazing.

Vibhu [00:27:03]: The first couple days.

Swyx [00:27:04]: But like, it. Like, they can clone a Jev API, which is honestly structured outputs

Nikunj Handa [00:27:09]: Yeah

Swyx [00:27:09]: Which OpenAI was first to.

Nikunj Handa [00:27:10]: Yeah.

Swyx [00:27:11]: Right? So, like, I think let’s iron out for people what is a decision model, as far as

Nikunj Handa [00:27:16]: Yeah

Swyx [00:27:17]: As far as, like, what is important? It is not just latency. It’s not just structured output, right? Because I could just have Luna as it’- The decision model is priced the same as Luna, right?

Nikunj Handa [00:27:26]: Mm-hmm.

Swyx [00:27:27]: Have turned off reasoning and then have structured output. Do I have a Jev? you know, no, right? And that’s the

Nikunj Handa [00:27:33]: Yeah

Swyx [00:27:33]: That’s the real

Vibhu [00:27:34]: There’s a confidence there.

Swyx [00:27:35]: Yeah.

Nikunj Handa [00:27:36]: Yeah, totally. I think, the way that. So we haven’t trained, like, a new model for this.

Swyx [00:27:40]: Yeah.

Nikunj Handa [00:27:40]: We’re, like, building this purely on top of the same Luna weights that we have.

Swyx [00:27:44]: Oh.

Nikunj Handa [00:27:44]: So yeah. This is, like, really just Luna. And, on top of that, what you’re doing is you’re constraining. So, like, structured output’s a big part of it. you’re really optimizing the inference stack to, like, get very fast on TTFD. And because you can have multiple questions, what you do is, like, you basically run those in parallel,

Swyx [00:28:05]: As a batch.

Nikunj Handa [00:28:06]: Yeah. You run those in the-- as a batch. you-- All sorts of, like, inference techniques people are working on to try to make it as fast as possible. But I’d say, like, at least our implementation of it at the start and this first version is, like, zero-shotting this on top of Luna, to see how it goes. And obviously, you wanna, like, put it out there. Like, this is OpenAI’s, like, classic iterative deployment thing. Put it out there, see what people think, and then, like, we’ll make more model improvements, as needed. so yeah. That’s, the decisions API.

Swyx [00:28:38]: Yeah. And, obviously as a benefit, you have vision. They don’t have vision, right?

Nikunj Handa [00:28:42]: That’s true.

Swyx [00:28:42]: Obviously, Jev’s comes with

Nikunj Handa [00:28:43]: Yeah. Like, we get it for free with Luna. Yeah.

Swyx [00:28:45]: Yeah. I do think that, like, you know, some of the innovations, it sounds like, it’s still to come if it’s still the same Luna weights, which is, like, the confidence stuff, like, the in calibration is something that we’ve talked about on the podcast with, benchmarking calibration. ‘Cause basically, the whole point is that RLHF kind of collapses you towards what you want to hear.

Calibration, Architecture, and the Open Research Questions

Nikunj Handa [00:29:03]: Yeah.

Swyx [00:29:03]: But, like, not actually, like, what the amount of confidence is.

Nikunj Handa [00:29:06]: Yeah. Yeah, totally. I’m eager to see how it pans out. Maybe there’s, like, gonna be. These are gonna be, like, the key areas where we may have to, like, hill climb

Swyx [00:29:15]: Yeah

Nikunj Handa [00:29:15]: With the, with the future model release. But, yeah.

Swyx [00:29:18]: And then architecture-wise, the other thing that’s in the debate, obviously, you-- Nobody knows because Jev doesn’t talk about it, but the two speculations are, one, maybe diffusion model instead of autoregressive.

Nikunj Handa [00:29:28]: Mm-hmm.

Swyx [00:29:29]: But you are able to achieve the parallel, generation in your way. And then the other one is some mech interp type thing

Nikunj Handa [00:29:37]: Mm-hmm

Swyx [00:29:37]: That you’re, like, analyzing the activations and then just outputting

Nikunj Handa [00:29:40]: That would be cool

Swyx [00:29:41]: The weights.

Nikunj Handa [00:29:42]: Yeah.

Swyx [00:29:42]: Which, like, you guys have all done the research on this. People have speculated.

Vibhu [00:29:45]: There have been demos on

Swyx [00:29:46]: Yeah

Vibhu [00:29:46]: Both of these as well. I think Gemini shared a Gemini diffusion, Gemma diffusion on a Jev-style output.

Nikunj Handa [00:29:53]: Oh, sick.

Vibhu [00:29:53]: And, interp people have also, you know, pulled out interp from a middle layer, but this is all speculation.

Swyx [00:29:59]: It’s just like, what are you trying to aim for, right? Because you can achieve the API. Everyone can achieve the API. It’s actually pretty trivial. But, like, then there’s the speed, then there’s the accuracy, then there’s the other calibration features.

Nikunj Handa [00:30:11]: Mm-hmm.

Swyx [00:30:11]: I don’t know what else.

Nikunj Handa [00:30:13]: Yeah. Yeah. No, totally. It’s so cool that this, like, whole space has been kicked off now and people are gonna do so much cool stuff and everyone’s gonna learn from each other. And, yeah, I’m excited about it.

What Developers Should Build Next

Vibhu [00:30:24]: I feel like being on the platform team, a lot of your job is to empower builders.

Nikunj Handa [00:30:27]: Mm-hmm.

Vibhu [00:30:28]: What do you think people should build with decisions API and also Computer Use agents? Any stuff that you’ve- been building with internally that you think really opens up after the new change?

Nikunj Handa [00:30:39]: Yeah. okay, let’s think. decisions API, use cases internally have been pretty obvious. Like, the user ops team was, like, jumping on it. We were like, “We gotta classify all of our support tickets.” what else came up? obviously, there were, like, the really cool GPT Live demos. I’m sure, like, the Codex app team might, like, pick this up and try to do something cool with it. So, you know, like, this whole thing started, like, a week ago, so it’s, like, very early and

Swyx [00:31:06]: Oh, one week.

Nikunj Handa [00:31:07]: We’re excited. Yeah. Yeah, pretty much.

Vibhu [00:31:08]: There was a big push in, evals, LLM as a judge having really low latency there.

Nikunj Handa [00:31:13]: Right. Yeah. That’ll be interesting to see. and then, with the Agents API, we have-- we’re basically, like, having a bunch of first-party products, like, at OpenAI built fully on top of it. we’ve had the Codex security stuff that just went out that’s fully built on top of, the Agents API. We have, sort of the-- we- we are having, like, a meetings type of thing launching today.

Agents API and OpenAI’s First-Party Products

Swyx [00:31:40]: Mm-hmm.

Nikunj Handa [00:31:40]: I think there was, like, a demo. do you remember, like, the plugin extensions when Sam was showing it? There was, like, a demo for, like, you’re in a calendar, you can sort of, like, have your meeting notes

Swyx [00:31:51]: Like, drop into a single

Nikunj Handa [00:31:52]: Flow into like your space

Swyx [00:31:52]: Like, Google Docs type thing.

Nikunj Handa [00:31:53]: Yeah.

Swyx [00:31:54]: Right?

Nikunj Handa [00:31:54]: And so the-- all of that stuff is, like, fully built on top of, the Agents API. and yeah, I’m, like, just excited to see. Like, we’re just getting this out, and let’s see what people build on top of it.

Vibhu [00:32:04]: I think you showed it off very well. The whole edit spaces, pages, collaborate, add in your dot. Like, that’s a lot, so

Nikunj Handa [00:32:12]: Yeah

Vibhu [00:32:12]: There’s a lot of inspiration people can go to.

Nikunj Handa [00:32:14]: Yeah. All possible with Astra, you know. Like, thing- things just move so fast now. Like

Swyx [00:32:19]: Yeah

Nikunj Handa [00:32:19]: People go from idea to execution so quickly, it’s amazing.

Swyx [00:32:23]: Is there something that you want, people to focus on to give you feedback? Like, what-- like, you know, maybe you’re just putting this out there and you want-- and there’s, like, a fork in the road and you want developers to help you decide.

Responses API Performance and Long-Lived Caching

Nikunj Handa [00:32:35]: So I think Agents API and decisions API, they are like, these are our newest products. Would love, like, any and all feedback on that to figure out where to take them. I think, over here, we’re, like, very open on Responses API, which is sort of like our workhorse over here. like, really focused on performance right now, and the performance comes in, like, two main ways. first is just, like, latency. We’ve been, like, rewriting the whole Responses API stack to, like, make it as fast as possible from a TTFT perspective, DVD perspective. So there’s like-- that, like, continues to be, like, a main area of focus for us. The second thing we’ve been trying to do is, like, really go deep on caching, particularly with these, like, personal agents that are, you know, like, basically, like, a single thread that just goes on and on forever. We’ve been, trying to, like, really up our game on caching. We provide now guarantees of, like, cache hits within, like, 30 minutes. We’re actually, like, we-- for one of our users, we just launched, like, a much longer cache window. So we have, like, a 12-hour caching guarantee, that we offer so that you have, like, guaranteed cache hits for

Swyx [00:33:40]: Is that a public API?

Nikunj Handa [00:33:42]: Not yet. That’s in preview.

Nikunj Handa [00:33:43]: We’re gonna, like, try to get that out to everyone as soon as possible. But, like, just pay a little bit more for the cache write, and we, like, guarantee, like, cache reads for, like, a much longer period. So even if, like, your instinct thread, for example, like, you just, like, do something on it and then come back to it, like, three to four hours later, you- you’re still getting the caching performance out of it. And launched

Vibhu [00:34:04]: And you cut the cost there quite a bit too, right, with the new model?

Nikunj Handa [00:34:07]: Oh, yeah. That’s right.

Vibhu [00:34:08]: Like, 25% cheaper, so

Nikunj Handa [00:34:08]: Yeah, with, like, driving down cache reads, yeah.

Cache Pre-Warming and Cost-Efficient Agent Threads

Vibhu [00:34:10]: For builders, they should implement

Nikunj Handa [00:34:13]: Yeah

Vibhu [00:34:13]: Because it’s significantly cheaper.

Nikunj Handa [00:34:14]: Yeah. Yeah. Just, like, building your apps with, like, to be very cache aware and sort of, like, use our prompt diagnostics or cache diagnostics tool to figure out, like, where things are dropping off. And, so the caching part is, like, really important. yeah, I also wanted to talk about pre-warming. We have that in the API now. So, like, if you know that, “Hey, I’m gonna get a cache,” like-- sorry, “I’m gonna get this prompt. I just wanna, like, pre-warm the cache, pay, like, the cache write fee right now, and then, like, have it sort of ready to go for the next 30 minutes for whenever.”

Swyx [00:34:49]: And it can spawn many instances of that thread.

Nikunj Handa [00:34:51]: Exactly, yeah.

Swyx [00:34:52]: Yeah.

Nikunj Handa [00:34:52]: You can just keep going and have

Swyx [00:34:54]: Yeah, just keep messing with the prompt there

Nikunj Handa [00:34:55]: Tons and tons of that. and so, yeah, like, I’m very excited about getting feedback on, like, the low-level performance things that we can keep making Responses API the most performant and reliable way to, like, build on top of an LLM. And then you basically have our, like, new products where I’m just looking for, like, any and all feedback.

Swyx [00:35:15]: Yeah, just use it, right?

Nikunj Handa [00:35:16]: So yeah, just use

Swyx [00:35:16]: Tell us what to

Nikunj Handa [00:35:17]: Yeah. Define our roadmap for us, please. So yeah.

Swyx [00:35:20]: I think for me, the caching thing, great, right? Like, obviously very needed. But at the end of the day, you’re still bumping up against a million-token context

Compaction and Managing Million-Token Contexts

Nikunj Handa [00:35:28]: Mm-hmm

Swyx [00:35:28]: And that’s probably not gonna change for the foreseeable future.

Nikunj Handa [00:35:31]: Mm-hmm.

Swyx [00:35:31]: Like, you still need good compression.

Nikunj Handa [00:35:33]: Yeah.

Swyx [00:35:33]: What is the best practice there?

Nikunj Handa [00:35:34]: Yeah. Yeah, totally. so firstly, OpenAI has its own, like, proprietary compression, comp

Swyx [00:35:40]: Which is in

Nikunj Handa [00:35:41]: Compaction.

Vibhu [00:35:42]: Compaction.

Swyx [00:35:42]: It’s in the agents.

Vibhu [00:35:43]: It’s in the API.

Nikunj Handa [00:35:43]: Yes.

Vibhu [00:35:43]: Agents API.

Nikunj Handa [00:35:44]: Yeah.

Swyx [00:35:44]: You decide for us, right?

Nikunj Handa [00:35:45]: Yeah, exactly. So in the Agents API, it comes built into the harness. and if you’re in Responses API, there’s, like, two ways of doing it. One is what we call server-side compaction, which is you basically tell Responses API that if you ever hit this threshold of tokens, just auto-compact it and, like, go back, or sorry, like, reduce the context, being used. And the second way is, like, /compact, which is, like, if you want full control. So you can, like, /compact at any time

Swyx [00:36:15]: I hear you

Nikunj Handa [00:36:15]: Have your own logic on when to, like

Swyx [00:36:17]: It’s not AGI.

Nikunj Handa [00:36:18]: It.

Swyx [00:36:18]: It’s not AGI.

Nikunj Handa [00:36:19]: Yeah. Yeah.

Swyx [00:36:20]: Yeah. But it, I mean

Nikunj Handa [00:36:20]: Yeah

Swyx [00:36:20]: It is the manual override.

Nikunj Handa [00:36:21]: Yeah, it is the manual way. And like, I don’t know, but a lot of the big coding agents like to do it manually. I mean, like, if you look at the Codex implementation of it in the Code- open source Codex harness, you can see that they use /compact and do it. and, there’s also, like, new, by the way, new compaction techniques that we are working on. Some of them you will be able to see in the Codex harness. Like, it’s already implemented in the Codex harness. And so, they’re like some file-based, systems that we are, like, experimenting with. So yeah, lots of cool stuff going on around in compaction as well.

Swyx [00:36:57]: Cool. we are running out of time.

Nikunj Handa [00:36:59]: Okay.

Swyx [00:36:59]: I think you’ve talked about, a lot about performance and talked a lot about, the new APIs that you’re launching. Can you give us any other hints as to things that you’re interested in as far as the future of the platform is concerned?

Higher-Level Platform Primitives and the AI Cloud

Nikunj Handa [00:37:13]: We’re obviously like very low level. Like, I used to work at Stripe before this, and, at Stripe a lot of the game was like building these higher level primitives and products on top of like the core payments primitives. and, I’m always like curious about what the best way of doing that is in AI. And I think we’ve had a couple of attempts at that. We like had launched assistance API like way back in the day, and like wasn’t really the right fit. We were sort of like going off with this like Agents API, and, it gives you the codex harness, but like where’s like the, what’s the right amount of flexibility to give in that? That’s like an open question. Like how should we like have memory walls and like all of these like higher level like API objects to take away, also like to abstract away more, like storage concepts. Like this is like a whole, like, there’s a whole space that I’m like very curious about figuring out how we design. I think a lot of things in AI are just have a low-level API primitive and see an example harness and go and have your coding agent implement that. But how much of that do we build into the API is like a constant question that I’m thinking about.

Swyx [00:38:24]: Yeah.

Nikunj Handa [00:38:24]: So I don’t know if folks have thoughts on that. If anyone has ideas, it would be super interesting to hear.

Swyx [00:38:30]: Yeah. The analogy I always bring back to, and we’ll end there, is, you’re building an AI cloud, right?

Nikunj Handa [00:38:35]: Mm-hmm.

Swyx [00:38:35]: Like, which is, something that, Sam said a year ago

Nikunj Handa [00:38:38]: Mm-hmm

Swyx [00:38:38]: Where, and you’re, it’s almost like you’re kind of doing the AWS invention and you have to do, okay, this is EC2

Nikunj Handa [00:38:45]: Yeah

Swyx [00:38:45]: And this is S3, and this is like. But you’re doing the AI-native versions of each of these.

Vibhu [00:38:48]: There are a lot of analogies, so you’re pre-warming caches for stuff that you know will be

Nikunj Handa [00:38:53]: Yeah.

Vibhu [00:38:53]: And it’s nice that it’s all exposed to builders

Closing

Nikunj Handa [00:38:56]: Mm-hmm

Vibhu [00:38:56]: ‘cause it just opens up ways that you can build new things.

Nikunj Handa [00:38:59]: Yeah, absolutely.

Swyx [00:39:00]: Okay.

Vibhu [00:39:00]: Awesome. Well

Swyx [00:39:01]: That’s everything.

Nikunj Handa [00:39:01]: Thank you, guys.

Vibhu [00:39:02]: Thank you.

Nikunj Handa [00:39:02]: Yeah.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit https://www.latent.space/subscribe

We are excited to have Anthropic share their latest AI x Finance work at AI Engineer New York, coming up in 2 weeks!

In case you’ve been under a rock, here’s a non-exhaustive list of what Anthropic has been shipping since closing the largest fundraise of all time in May at $47B ARR:

* June: Launched Claude Tag and Sonnet 5 and Fable 5

* July: Opus 5, /checkup. crossed $65B ARR

* Last month: Fable/Mythos 5.1, and EFS (upcoming pod)

* IPO target $2T, end 2026 ARR estimated $100B

* Cowork/chat merged before did

* Claude Mods

* Dario endorses the same Pacing the Frontier message cosigned by all labs

* Last week: Opus 5.5, Plugins portal, Cloud Sessions/Claude Projects

* Today: Sonnet 5.5!

Today’s episode should catch you up, with Thariq Shihipar, the explainer-king of Anthropic, who we last caught up on Fable launch day with The Field Guide to Fable:

The Future of Mutable Software

Pay special attention to Claude Mods (especially the cheatsheet):

In general this is also the inverse of the other viral tweet from Thariq:

Cloud Brain, Local Hands

And give a try to Claude Projects:

The “hands” terminology is not just an analogy for the local/cloud paradigm that is being built up at frontier coding agent companies like Cognition, but is ALSO particularly relevant to the safety systems discussions that we’ll be discussing with Anthropic in an upcoming episode as they prepare to pace to frontier with responsible AI deployment.

For those who want Thariq’s writing tips we teased at the start of the pod, watch the full video here:

From the rapid rise of Claude Code to a future where agents can rewrite their own harnesses, collaborate across teams, and operate across cloud and local environments, the way we build software is changing extraordinarily fast. In this episode, Anthropic’s Thariq Shihipar joins swyx and Vibhu to unpack how power users are actually working with Claude Code today, why prompting remains a high-skill discipline, and where Anthropic thinks the agent harness is headed next.

We go deep on Claude Code’s evolving interface: Ask User Question and elicitation, artifacts as persistent generative interfaces, Claude Tag for multiplayer agent workflows, Projects, model effort, implementation notes, and the new Claude Mods system for customizing the harness itself. Thariq explains why Claude.md may eventually disappear, why the smartest model could also become the cheapest model for many tasks, and why mutable software could become a new paradigm for how applications are built and customized.

The conversation then turns to agent security and Anthropic’s “Pacing the Frontier” argument. Thariq walks through recent incidents where agents discovered unexpected ways to communicate, exploit infrastructure, reverse-engineer benchmark scorers, and chain vulnerabilities together. We discuss sandboxing, prompt injection, autonomous agents, interpretability, constitutional classifiers, probes, fallbacks, Auto Mode, and why securing increasingly capable agents may become one of the defining engineering problems of the next few years.

We discuss:

* Why agentic coding went from controversial to the default in less than a year

* Why prompting is still one of the highest-leverage skills for working with Claude Code

* How expert users build a mental model of Claude and what it can reliably one-shot

* Why discovering your “unknown unknowns” matters more as agents become more capable

* Artifacts as persistent, generative interfaces between humans and agents

* How Claude could split into a cloud-based “brain,” local or remote “hands,” and dynamic interfaces

* Claude Tag, Projects, and multiplayer agents and how collaborative agent workflows could evolve

* Why spending more time on the initial prompt can dramatically reduce wasted agent work

* When to use low, medium, high, or max effort for different engineering tasks

* Why frontier models may eventually outperform smaller models on both intelligence and token efficiency

* Why implementation notes can expose decisions the model considered but chose not to make

* Why Claude.md may eventually disappear — and why starting without one can sometimes be better

* Claude Mods: customizing the execution loop, UI, subagents, routing, and behavior of Claude Code

* Model routers, forked agents, and supervisor agents that automatically improve agent workflows

* Why Claude Mods may be an early preview of “mutable software”

* The bitter lesson of harness engineering and why agent architectures go out of date so quickly

* How Claude Tag is becoming an organizational harness for multiplayer work

* Why giving agents access to company data creates an enormous new security surface

* The Exploit-Bench incident where agents discovered ways to communicate and collaborate

* Why agents hacked Hugging Face for scorer code rather than benchmark answers

* How agents chained sandbox and infrastructure vulnerabilities in unexpected ways

* Why increasingly capable agents make traditional security assumptions harder to maintain

* The argument behind Anthropic’s “Pacing the Frontier” proposal

* Why software engineers are increasingly doing two jobs: engineering and keeping up with AI

* Constitutional classifiers, probes, and fallbacks and what interpretability looks like in production

* How Auto Mode checks whether an agent’s actions actually match the user’s permissions

* Why Thariq can see serious AI risks while still having a relatively low p(doom)

Thariq Shihipar

* X: https://x.com/trq212

* LinkedIn: https://www.linkedin.com/in/thariqshihipar/

Timestamps

00:00:00 Introduction

00:04:12 Ask User Question and the Future of Agent Interfaces

00:08:29 Artifacts, Projects, and Multiplayer Agents

00:15:37 Prompting as the Core Claude Code Skill

00:21:52 Context, Effort, and Smarter Model Usage

00:28:10 Is Claude.md Going Away?

00:32:49 Claude Mods: Customizing the Claude Code Harness

00:36:35 Model Routing and the Rise of Mutable Software

00:44:40 The Bitter Lesson of Harness Engineering

00:50:49 Claude Tag as an Organizational Harness

00:55:59 Pacing the Frontier and Autonomous Agent Security

00:58:22 Agents Hack Hugging Face for the Scorer

01:05:34 What Happens When Agents Need More Compute?

01:10:32 AI Coding Is Changing Faster Than Engineers Can Keep Up

01:17:17 Probes, Fallbacks, Interpretability, and Auto Mode

01:28:32 AI Risk, p(doom), and Closing Thoughts

Transcript

Introduction: Life at Anthropic and the Pace of Change

Swyx [00:00:00]: We’re here in the studio with our friend Thariq from Anthropic, and I guess generally the Claude Code, I-- there’s, there’s so much, merging of boundaries and you’ve been so on top of everything since you joined Anthropic. You have been early to Claude Code itself, but then also, and you’ve told that story in other podcasts, and you’ve also been talking about seeing like an agent. Most recently you did the top AIE World Tour talk, Field Guide to Fable, which obviously you guys launched Fable, so that was-- that’s cheating. And mostly you most recently also launching Claude Tag, and we’re also gonna be talking about Pacing the Frontier. There’s a lot going on in Anthropic. I guess top of the question is, what’s it like being at Anthropic when there’s so much going on?

Thariq Shihipar [00:00:48]: I think that It is, like. I think you can get whiplash sometimes. I think, like, going. When I joined Anthropic, I joined because of Claude Code. Like Claude Code had just come out and I was like, “This is so good.” And Opus 4 to me was like just, I could not imagine, like, how good it was? And that was, like, a real moment for me. But I was, like, trying to convince, like, my startup friends to use agentic coding, and they’re like, “Oh, no, like, our engineers don’t think it’s good enough,” or something. And I was like, “That’s insane.” and now you, like, fast-forward, 12 months, less, and, like, it’s just like, yeah, the default way that everyone codes, right? And I think that, like, just having to go from, like, selling it to, like, now, teaching people how to be. make the most use of it and be more efficient and things like that is just like a big, like big change. And, yeah, I think, like, it’s just hard to stay on top of everything as a human? Like, I think things happen so fast and like

Swyx [00:01:51]: You just throw more agents at it.

Thariq Shihipar [00:01:52]: Yeah, like that’s like the agentic stuff scales much better than the, like, human stuff where it’s like, oh, like, there are three things happening right now and, like, they’re all emergencies and, like, how do you, like, respond to it? Yeah.

Teaching People to Use Claude Code

Vibhu [00:02:05]: What do you split your time on? You do a lot of technical writing, engineering work.

Thariq Shihipar [00:02:10]: Yeah, so I think that, like, when I joined the Claude Code team, I wanted to teach people how to use Claude Code and I think that, like, that has been something that, like, I thought, like, maybe I would spend a little bit of time on it or, like, I’d, like, do. I was spending some time on the agent SDK first, and I wasn’t exactly sure, like, how the bitter lesson would go, when it comes to, like, harnesses, right? Like, I think sometimes we were like, “Oh, like, what’s after Claude Code?”? And so initially I was like, I just wanna teach people how to use Claude Code and make it easier to use Claude Code. And I think that has just, like, as the harnesses have gotten better and better, that’s like the dominant problem now is, like, how do you use the agents, right? Like, it’s like such a high skill expression thing. So I do that and then I do engineering work. I give talks, but I think, like, when I’m doing engineering work, my goal is to take that feedback that we get from users and also, like, then be able to talk about, like, hey, how to use Claude Code to do engineering. So there’s like a good loop there. Yeah.

Swyx [00:03:07]: Yeah. I’ll-- For listeners, we’ll attach, the talk that you did with Sarah for the Dev Writers, meetup

Thariq Shihipar [00:03:13]: Oh, yeah

Swyx [00:03:13]: Which we talked a little bit about, well, first you do the work and then you talk about the work.

Thariq Shihipar [00:03:16]: Right.

Swyx [00:03:16]: Something like that.

Thariq Shihipar [00:03:17]: Yeah.

Swyx [00:03:17]: It’s sow and reap or

Thariq Shihipar [00:03:19]: Yeah, reap and. Sow and reap.

Swyx [00:03:21]: Something like that. Something like that. Yeah, so, and then just to preview a little bit, we are gonna talk about the evolution of the harness. It has come a long way from just being a CLI. We’re gonna talk about, Claude Mods, which is starting to leak today, because you couldn’t keep it secret.

Thariq Shihipar [00:03:36]: Yeah. yeah.

Swyx [00:03:39]: Yeah, there’s, there’s a lot, there. I think you started off with, like, adding ask user question tool, which people love and hate.

Thariq Shihipar [00:03:48]: Yeah.

Swyx [00:03:48]: Like, I thought it was, like, very innovative, and then now I have, like, my own version. You have your Interview Me version.

Thariq Shihipar [00:03:55]: Yeah.

Swyx [00:03:56]: And, yeah, everyone just has, like, their own stuff. And, like, it no longer matters ‘cause now you’re supposed to, write prompts that create other prompts and loops and all these things.

Ask User Question and Human-Agent Interaction

Thariq Shihipar [00:04:05]: Sure, yeah.

Swyx [00:04:06]: So what’s the state of the art, today? Like, what are people. what are you, like, telling people to do today?

Thariq Shihipar [00:04:12]: Yeah, ask user question was the first time that the model was good at elicitation. I think this was, like, an emergent behavior that I, like, wanted to see if the models could do. I have, like a human-computer interaction background, so I, like, did that in undergrad and grad school. And so this was like. I think it’s like human-agent interaction to me, like, trying to figure out, like, how can the agent communicate with you and extract, the requirements, right? I think that, like, one of the things about, like, that’s difficult as Claude Code has gone broader and broader is that everyone has, like, their own way of using it, and it’s very hard to, like, change the default behavior. So for example, like, if someone asks Claude Code to do something,

Thariq Shihipar [00:04:59]: Sometimes they just want them to do the work, ‘cause they’re, like, maybe a very good prompter, and sometimes they want. like, are not good at prompting? And you need. like, the agent needs to, like, clarify? And so that’s, like, a good split. Like, and the ask you the question tool like, splits along that side where, like, are-- do you feel like you’re good enough to instruct the agent as it is, or is the agent able to, like. does the agent need to, like, pull out more requirements and, like, collaborate with you more and really understand your preferences?

Thariq Shihipar [00:05:27]: I, on the whole, believe that pretty much everyone is more on the latter than the former, that they, like, have more ambiguity and they know less than they want, than they, like, think they know about the problem. but, like, it’s like a interface design problem to make that easy? And so, like, if you’re designing a problem, like, or if you’re going through a problem, like, things like what’s the schema or, like, what’s the call stack and things like that are really important. like, the details in the design are important. Ideally, you want to figure out some of these, like, hard problems ahead of time before starting implementation. And yeah, that’s why they call, like, unknowns, right? And so I think that this will forever be, like, a skill in agentic coding is, like, figuring out your unknowns. So, like, because even if the model is, like, super intelligent- It, like, needs to know what you want? And, like, you have preferences. like, you need to like, pull the, pull that out. and so that’s, like, I think how I’m, what I’m pushing. the question then is, like, how does the agent interact with you? And I think that has been HTML, has been, like, the big way of doing that. And we’ve recently added artifacts, right? And artifacts, I think we’ve done a bad job of, like, or, like, I’ve done a bad job of, like, explaining how to use them fully. We have a lot of property capabilities. They have a database associated with them? And so every artifact can store and write persistent data. They can, like, feed back into Claude? And so, like, one thing that, like, people are not doing yet that I’m trying to, like, encourage is, like, this idea of a dashboard artifact. So you have, like, Claude working on a project long-term. Maybe it’s like a kanban or something. it can store that kanban data in its database. Multiple Claudes can access that data via, like, the artifact MCP, and, like, that artifact can, like, talk to those Claudes as well. And so, like, the. We’re building the primitives for you to be able to have this, like, generative interface via artifacts that will, like, let you surface more of that rich detail from the agents. And I think that, like, almost everything with agents right now is, like, this problem of, like, you think what you want, but you don’t really know what you want, and, like, the agents need a lot of detail, and collaborating with them in the loop is really important. And so artifacts are, like, the, like, way that we’re trying to evolve there. But there’s a lot of work to do because it’s so much more complicated than, like, a multiple-choice question? there’s a lot more, like, detail in terms of, like, diagrams and code snippets and schemas or, like, whatever it is for that problem. But, like, artifacts is, like, the mo-more AGI-pilled way of, like, doing ask user question. So yeah.

Artifacts as the Interface to the Harness

Swyx [00:08:15]: I think one thing that’s unclear to me about these, the artifact stuff is, like, what feedback should go in through the artifact and what feedback should go through a Claude, a chat? Because the more AGI-pilled one is to just feed everything to the Claude.

Thariq Shihipar [00:08:29]: I think the more AGI-pilled one is to go through the artifact. Like, and I think that, like, we imagine in the limit, I think that artifacts will be your interface into the harness? You can, like, comment on this, like, live, like, document of your plan, of the work. you can see maybe, like, multiple agents and different agents are doing this, and that artifact is built for the current work that you’re doing, right? And so, like, each one has, like, slightly different. I think we’re still, like, getting there from, like, an infrastructure perspective. But yeah, I think, like, on-the-fly interface for your harness is probably where things are headed.

Vibhu [00:09:03]: Is there a version of it that’s an abstraction from CLI or chat and you. Because right now, a lot of it is, okay, you’re interfacing with Claude Code, you’re having HTML given back for a mockup. It’s pretty rich. There’s diagrams. Artifacts are ways to connect these together. Why not just do everything that way?

Separating Brain, Hands, and Surface UI

Thariq Shihipar [00:09:22]: Then it becomes, like, separating out, like, where is the inference happening? Where is the intelligence happening? Where is the work happening? like, I think this is like, difference between, like, or, like, some of the distinction between local and cloud, right? And so, I think right now, if you use Claude Code, it’s, like, local and, like, you can spin off remote control, for example, to get some cloud behavior, or you can spin off Claude Code in the cloud, right? We’re moving towards a place where instead of Claudes, like, you message a local Claude, it starts a session locally and it executes, to more like you have a Claude that you message that’s in the cloud that’s running. it can run, like, local, or, like, cloud sessions. This is how Claude Tag works. But, like, over time, we’ll add, like, local hands as well. And so, like, local hands will be the ability for that agent to access your computer if it’s online, and be able to, like, work there. And so it can spin off many different subagents. It can, like, commu- those subagents can communicate with each other, and that’s where the artifact comes in to display all of that work. So you can imagine, like, the. You’re separating out these things. So there’s, like, the surface UI display that’s an artifact and hosted somewhere and has a database and everything. There is the inference intelligence, right, that’s happening on the cloud, and you don’t have to worry about shutting off your computer or whatever, right? and then there’s the, like, hands. Like, and it can be local, it can be in, like, a remote sandbox or wherever you need your work to be done. That’s like unpackaging, like, the Claude Code experience right now where, like, right now it all happens in one place, right? So.

Multiplayer Agents, Claude Tag, and Projects

Vibhu [00:11:00]: How do you see, like, the multiplayer side of that? So say teams want to work in this way. Right now it’s very individual, but how do you see the future of multiplayer? Like, right now, I guess there’s Claude Tag, which is a version, but.

Thariq Shihipar [00:11:12]: We’re launching projects. And so projects is the, like, this abstraction that’s like Claude Tag, but on our Claude products, right? So you can message it and, like, it will do the Claude Tag-like stuff, like spinning off subagents. So We think with multiplayer. Like, Claude Tag is, like, a little bit more native multiplayer because it’s just, like, in your Slack and the permissions are all figured out and stuff like that. But I do think multiplayer is, like, an important part of the story and, like, that will need to get tied together more. Like, you can imagine how complicated it gets when you’re like, oh, you have hands, but now you have other hands in other people’s computers too, and, like, you need to, like, permission them or, like, you have, like, your MCP and someone else’s MCP, and how do you figure out how to use them, right? It gets, like, quite complicated. And Claude Tag does a good job of, like, sanding down all of these issues, right? So that, like, when you have, yeah, Google Docs, how does it access Google Docs, right? Like, it accesses through the shared Claude MCP, or it can access through your local credentials as well if it doesn’t have access. But yeah, I think Claude Tag is our multiplayer, product, and it’s really useful for these, like, things that are inherently multiplayer. Like, okay, like on-call, for example, incidents are inherently multiplayer. You want to tag Claude, you want multiple people to log in, you want it to be able to find context. I think whenever I’m, like, working on something and I want, like, privacy or security or, like, I want other people to review it’s really nice to, like. I’ll have a channel per project and I’ll, like, at legal, for example, be like, “Hey, like, I want to ship this. Can you, like.” Like, here’s. Like Claude knows everything, just chat with it. And that way legal gets precise answers, on like what exactly is shipping into the code, and I don’t need to be in the loop, right? So I think like multiplayer is getting like more and more like, yeah, everyone can participate with Claude. I think Claude Tag is like that product and like projects will start off single player and will like, expand.

Swyx [00:13:14]: I think there’s a question about like maybe dual questions about identity and the unit of isolation.

Identity, Permissions, and Isolation

Thariq Shihipar [00:13:20]: Yeah.

Swyx [00:13:20]: Claude Tag, you specifically chose to make it its own identity

Thariq Shihipar [00:13:26]: Yes.

Swyx [00:13:26]: Which is like, a controversial choice. There’s, there’s other ways to do it.

Thariq Shihipar [00:13:30]: Yeah.

Swyx [00:13:30]: Claude Projects probably it sounds like, if it’s anything like ChatGPT Projects, it is, the isolation is that artifacts, that cloud instance, everyone’s collaborating on this. It’ll. It sounds like, it should be like if you’re, if you’re collaborating with legal on a thing, like that channel should be a project, right? Like it’s not yet

Thariq Shihipar [00:13:50]: Yes.

Swyx [00:13:50]: But it. that’s the natural next step.

Thariq Shihipar [00:13:53]: Yeah, like I think in Claude Tag, it’s effectively. Like Claude Tag, you have to do your own arrangement. And so Claude Tag, yeah, each channel is like you can name it as you want, and I name

Swyx [00:14:04]: Yeah.

Thariq Shihipar [00:14:04]: Like each feature

Swyx [00:14:06]: Yeah.

Thariq Shihipar [00:14:07]: As a channel.

Swyx [00:14:07]: And, but I think like there is some trans- like it’s unclear when there is transference, because let’s say it is. if you have a coworker

Thariq Shihipar [00:14:14]: Yeah.

Swyx [00:14:14]: Who is tagging on all these things, yes, there is transfer

Thariq Shihipar [00:14:16]: Yeah.

Swyx [00:14:16]: Because it’s the same person. but with Claude, it’s unclear if it’s like necessarily like, well, no, you don’t know any of. you don’t know about the other stuff. You should only use this stuff.

Thariq Shihipar [00:14:25]: It’s like the tip of the iceberg meme, right, where you can like. This is what we spend so much time on

Swyx [00:14:31]: Yeah.

Thariq Shihipar [00:14:31]: Is like there is like infinite surface area of like, okay, you want Claudes to. Not infinite, but like there’s like surface area, a lot of like, surface area to figure out of like permissions and visibility and like how can you let Claude operate as well as you can, as safely as you can? And obviously, this is very important to us because like security for our code base is very important. And so we’ve put a lot of time into this. Yeah, there’s so many like edge cases you can figure out where it’s like, oh, like, yeah, this Claude in this channel has different permissions, but it can message another channel, and can’t it exfiltrate data that way? Or like can you like. What if it uses your MCP and then messages someone else? Like there’s like so much, and we’ve like really put a lot of work into sanding it down.

Swyx [00:15:14]: Yeah. Lots of work. okay. Fable?

Fable and the Meta-Skill of Prompting

Vibhu [00:15:18]: Fable, you wrote two good articles. you’ve written many good articles

Thariq Shihipar [00:15:22]: Yeah.

Vibhu [00:15:22]: But on, Field Guide to Fable, Building Claude Code. I’m curious from what you’ve seen, is there any common patterns that you see in like top users at Anthropic externally? Like what are best practices for getting the most out of Claude Code?

Thariq Shihipar [00:15:37]: The like meta skill I say is like prompting is like very important? And like that. Like I think this is like not trivial to say because I think a lot of people are like, “Oh, prompting doesn’t matter. It’s just like I can just say a sentence and Claude will do it.” And I think prompting is really this like, this. It’s like public speaking, like, or writing or something, and for a specific audience, and that audience is Claude. And you need to like build a mental model of Claude and how it thinks and how it works, right? And so that’s like the most important skill in working with Claude Code is like having this mental model, right, of Claude and like what it can do well, what it can one-shot, what it can’t. And so many people when you see prompting, they’re just like, they’re short prompts, but they have such a good mental model of Claude and of like the code base and things like that like it’s effortless? But it’s like high skill ceiling. So like that work of like, spending a lot of time prompting and building mental models of how, and intuition for how the agents work is really important. And then I think like the next thing is like the unknown stuff we talked about earlier, where it’s like being able to find out like your, what you don’t know or what you haven’t written down, learning about like different things. I think as Claude can do more and more things, the likelihood of you doing something out of distribution for you and like you have low domain knowledge on is very high? And the more you can like learn the vocabulary to be able to prompt Claude, it becomes really important. And so like I think the most important unknowns are the unknown unknowns, where you’re like, I just like don’t even know that this exists, right? Yeah, exactly. I think that’s like a illustration of like the map and the territory, right, where you’re like, “Okay, this is my prompt,” and the territory is like the actual like work that the agent needs to do, right? And if you are like very precise, you can give more precise things, right? So like for example, in design, I’m not very precise. I’m not a designer, so I say like, “Give me like eight different mock-ups.” But if I was a designer, maybe I’d be like, “Oh, hey, here are some reference sites.” Like, “I want this type of font and this type of like look to it, and here’s like a few different components to like visualize. Here’s a Figma MC board to bring in,” like. And so you can just be so much more precise with that language. And if you’re not a designer, you just need to like try and learn the language or learn the unknown unknowns. And this is true of like everything, I think. Like the more, like you can work with Claude to learn like how things work, the better your prompting will be. I think another good example of this is like game design, like where a lot of people are like, “Oh, like I can vibe code a game now.” And they’re like, “It’s not fun.” And like it’s just like the thing about game design is like every one of these choices has like a lot of

Taste, Domain Knowledge, and Learning the Vocabulary

Swyx [00:18:25]: Variations.

Thariq Shihipar [00:18:25]: A lot of like craft to them. So it’s like, oh, okay, like when you’re making a flying game, the feel of the plane and the like, way it responds to your controls has a lot of like. Like, a game designer would spend like days on that. Do? and like

Swyx [00:18:44]: To me, that’s what taste is, right?

Swyx [00:18:45]: Like it is like from the possible space of one thousand mathematically valid answers

Thariq Shihipar [00:18:49]: Yeah.

Swyx [00:18:49]: Here’s the one that is the humans will like.

Thariq Shihipar [00:18:51]: Yes. Yeah.

Thariq Shihipar [00:18:52]: I think with taste, I’m like torn on this word ‘cause I think you’re right, but everyone has different definitions, and it sounds kind, sounds like low skill or like elitist almost, where you’re like, oh, like there are certain people with taste?

Swyx [00:19:06]: It’s like taste is what I call taste.

Thariq Shihipar [00:19:07]: Yeah, exactly.

Swyx [00:19:08]: And it’s like these guys don’t have taste.

Thariq Shihipar [00:19:09]: Yeah, exactly. Oh, like an engineer doesn’t have taste. Like I, the like founder, have taste.

Thariq Shihipar [00:19:14]: ? And I think that’s not true. Like I think like the engineers have a lot of taste for these particular like problems? And I think everyone has taste for particular problems. I think like Jason Liu, like say like in order to, yeah, have taste, you have to eat?

Thariq Shihipar [00:19:32]: And I really like that, where it’s like, okay, you have to like do a lot of things. You have to like iterate and figure out what you want, what you like, and, like build that like domain

Swyx [00:19:41]: Yes

Thariq Shihipar [00:19:41]: Domain vocabulary. And then when you’re prompting, you’re like synthesizing all of that for a product.

Swyx [00:19:46]: Isn’t it annoying when someone else says it better than you?

Swyx [00:19:48]: It’s just like, f**k, I have to quote this guy forever.

Vibhu [00:19:51]: Having to quote Jason Liu forever.

Vibhu [00:19:53]: He’s gonna love this.

Thariq Shihipar [00:19:55]: So I get prompts, more than that.

Vibhu [00:19:57]: And sometimes it’s not even that. Sometimes it’s just intuitive, right? Like you don’t realize you even want something till a model puts it out, and you’re like, “Oh, this just feels immediately better,” right?

Voice Prompting and Information Density

Thariq Shihipar [00:20:07]: Yeah, exactly.

Swyx [00:20:09]: One thing I go back and forth on is I feel like the way I prompt half the time, let’s say I use voice.

Swyx [00:20:16]: Did I say voice? Other people have voice. that is the opposite. That is just like me rambling for like two minutes Pressing down the function key and then let go, and then like hopefully it figures it out. And oftentimes it does.

Thariq Shihipar [00:20:26]: Yeah.

Swyx [00:20:26]: But it’s not as thoughtful as like a structured prompt with like Well-run communication as though it’s a PRD or a memo. Is that in line with how people do this? There’s like bimodal prompting where there’s some prompts where you spend a lot of time upfront and other prompts you just dash it off?

Thariq Shihipar [00:20:43]: I don’t think the voice is necessarily low. Like I think it’s like more like how much information is in the prompt. like the model can. Like you can and like add some sentences

Swyx [00:20:53]: Right

Thariq Shihipar [00:20:53]: And be like, “Oh, like I changed my mind,” like in the middle of the prompt, and it will be able to follow that perfectly? So I think the like actual format of the text is less important, but then like the ability to. Like how much information is in it, right? And I think for voice, a lot of times, going back to like human-agent interaction and like for a lot of people, it’s just way easier to talk than to like type? and I. If that gets more information out of you, like that’s better.

Vibhu [00:21:21]: At some level, it feels like just giving the model as much context

Thariq Shihipar [00:21:24]: Yes

Vibhu [00:21:24]: Over prompting before you kick off is a best practice. I don’t know. A lot of the times, like when I was first trying out Fable, I spend a solid 30 minutes like really crafting a long prompt. This, I think, is a response of models running for longer and longer, right? It’s still a little difficult to nudge them as they’re in like, in the loop, but I just like intuitively spend more time kicking off that first prompt and working with it a lot.

Spend More Upfront, Iterate Less

Thariq Shihipar [00:21:52]: My personal opinion is that if I was a software engineer, if I was like, just running my own startup, for example, I think I would mostly fit, stick to a max 20x? like maybe verification and so code review are like separate things. But I think like what I see a lot of times is people hit rate limits when they’re doing this like, oh, like it did a lot of work and you’re like, “Oh, I don’t like this.” Like, “Can you like undo this and redo it?” And then you’re like iterating on this like thing that the model could have done if you had like spent more upfront time or given it better context? And instead it’s like you’re like, “Nope, don’t like that design. Try this.” Or like, “You messed this up,” or something like that. And then that just eats up so much more of like, your usage. And so that’s like, I think maybe like a key like tip both for like efficiency as well, right? And yeah, I think like context, and not just like context on like what the goal is good, right? Like are you building a prototype or is it like a production thing? Like where can you spend compute or when, where can you not spend compute? Like I think you have to give the model permission or like not permission to do things sometimes where, like it doesn’t know intuitively how much you want to spend on this task, right? And you can use effort for this. So I did-- I’m working on a blog post about that where it’s like, if you want. For like we see that effort scales with the complexity of the task. So for security, effort gets like way more results. Like high effort versus like low effort gets, like changes the evals a lot. But for software engineering, it doesn’t change it a huge amount because effort is mostly spent on the verification and the like edge case testing and things like that. And so like being able to like give the model that guidance of like, “Hey, this problem is something that I think I want you to spend a lot of time verifying and edge case testing,”?

Effort, Model Choice, and Verification

Vibhu [00:23:43]: How about model in the mix? So, there’s Opus and Fable with effort.

Thariq Shihipar [00:23:47]: Yeah.

Vibhu [00:23:48]: There’s also Haiku in there.

Thariq Shihipar [00:23:49]: Yeah. It’s not quite true yet, but it’s very close where I think the frontier models will be Pareto dominant over like almost everything. like maybe. And sometimes I think Opus might be Pareto dominant. Do? Like I think depending on like how things, like shake out if it’s like a newer version of Opus. But I think that like increasingly it’s just going to be like the smart model is going to be able to like do the simple task for less tokens than the like the other models because of verification. With verification, in the limit, your model doesn’t need to verify, right? If it’s a perfect model, it just does the work once and it’s like, okay, like you, I did it? And increasingly with Fable, I’m like, I’m like, “Dude, you don’t need to spin up Chromium and screenshot all of these things.” Like I see it. Like you did it, right? And so a lot of the. At higher effort, you spend more of those tokens verifying. But if you’re working on simpler problems, and a lot of software engineering is like well, like in Fable, like low and medium stability, it can spend less tokens verifying. And as the models get smarter and smarter, they will just be able to like, “All right, done.”? Like, I can run the lint for sanity’s sake, but, like, I, like, know it lints? Like, you don’t even need to do that. And that will be so much more token efficient than, like, the smaller models. Yeah.

Swyx [00:25:15]: Is there a good, practice on our side that we can use to see if we’re using too much effort? Like, I freaking

Thariq Shihipar [00:25:23]: Yeah

Swyx [00:25:23]: Hate wasting time on that stuff.

Thariq Shihipar [00:25:24]: Yeah. I know what you mean. I think, like, so in this blog post, my rough distribution is, like, code review and security should be, like, high or max and, like, software engineering

Swyx [00:25:37]: You said recommend mix settings per domain.

Thariq Shihipar [00:25:37]: Yeah. I think, like, if you’re doing, like, UI or something like that, like low and medium, I think is you’re building, like, an API and you want to make sure, like, you cover enough edge cases? And so I think building, like I said, that mental model of, like, how things work across these distributions is, like, yeah, part of the job.

Implementation Notes and Decision Logs

Vibhu [00:25:56]: This is more intuition-driven or eval? Because I’m guessing this would change as you go.

Swyx [00:26:00]: He has evals.

Thariq Shihipar [00:26:01]: Yeah. So what I did in the blog post is I go over all of the terminal bench evals. So there are, like, 70 problems and I’m show that, like, okay, like, in the security problems it does more. and then I also, like, look at some of the transcripts just in terms of, like, how-- what does it answer, what does it forget or something. And a lot of times, this is another prompting tip I have, is, like, asking it to make decision notes or implementation notes because, in every eval problem that it faces, it thinks about the correct solution, and decides not to do it. it’s like, oh, like, here is the answer. What if I did this? And then it’s like, oh, probably not? and then keeps going. And this is, like, the majority of the failures, at, like, a higher max level. It’s very rare that the model just doesn’t know how to do something. If you just have these implementation notes, then you can review and you can be like, “Oh, I want you to do this thing that you didn’t do.” The models are getting better at surfacing that overall. Like, I see in the transcripts of Fable 5.1, like, when it does this output, it will call out its decision-making as well. but making this more explicit in the harness is better. And now we’re, allowing ways of you modifying the harness so you can, like, add some

Vibhu [00:27:23]: Ooh.

Thariq Shihipar [00:27:24]: Calculate with there. Yeah.

Swyx [00:27:25]: Yeah. So I do wanna call out two things that you mentioned that I think exist outside of prompting. One is like, let’s, let’s call it the prompt that is so important that it shouldn’t be in a prompt. It is in Claude.md or Agents.md

Thariq Shihipar [00:27:38]: Yeah

Swyx [00:27:38]: Which is like goals, right? Like your situation, your goals, the things that you want, the thing. and then second of all is the decision log or the experiment log or whatever log of traces that you might want to survive the current session to do those things. Those are, like, externalities that there’s no standard. There’s no-- It’s not like skills. It’s not like MCP. There’s no standard. It’s, it’s just like it’s a markdown file. first of all, is that right? Is Claude.md going away? You have a documented dislike of, Agents.md, but you’re gonna do it?

Claude.md, Agents.md, and Model-Specific Instructions

Thariq Shihipar [00:28:10]: Yeah. Okay. So Agents.md, yeah, like, we’re, we’re gonna do it. I think it’s just, like, different models are very different from each other? But I realize that it’s, like, such a pain to, like, maintain different ones? And yeah, like, as the models get better and better, the floor of how they accomplish the simpler task is better. And so I do think in the limit, Claude.md goes away, and maybe not even, like, that far. Like, I think, like, I think that right now it might be better to start a new project without a Claude.md.

Swyx [00:28:44]: Yes.

Thariq Shihipar [00:28:44]: I think that, like, maybe if you see very repeated failure modes, you add them to your Claude.md. The really tough thing is that this changes per model. And so, like, if you’ve added a bunch of failure modes or, like even

Swyx [00:28:57]: So you need Fable MD, you need Opus MD.

Thariq Shihipar [00:28:59]: Or well, even Fable 5.1 versus Fable 5.

Swyx [00:29:03]: Yeah.

Thariq Shihipar [00:29:03]: Like, it is annoying. Like, I’m not like,

Swyx [00:29:05]: Yeah

Thariq Shihipar [00:29:05]: Like, we don’t, like, do this on purpose? It’s just, like, how the models work, right? And so, like, maybe, like, Fable 5 had this, like, failure mode that Fable 5.1 doesn’t. And if you keep this context, this running log of a bunch of different failure modes, they will probably over constrain Claude? And so this is like. we just added evals plugins for skills.

Swyx [00:29:28]: Yeah.

Thariq Shihipar [00:29:29]: And so now you can eval if a skill is better. I think Daisy on our team did this. And so, yeah, this is like we’re trying to work on this. We know it’s, like, you still have to spend tokens on it and, like, it’s not, it’s not perfect, but it’s, like, we’re trying to help out with this problem.

Swyx [00:29:44]: And so, and as far as prompting goes, the one tip I wanna offer is, something I have told people a lot is sufficiently advanced prompting is indistinguishable from sufficiently advanced executive communication. So I’ve referred to-- This is an executive comms workshop from Heavybit that is the best I’ve ever seen in my career. And they teach this thing called the SCQA model. Just Google it. It’s a, it’s a thing. Like, people have done prompting for decades. It’s just called executive communication. It’s like when one person has to communicate to thousands of people down the org chart, this is what you do. so situation, complication, question and answer, is how you write the memo. but obviously sometimes you don’t have the answer, but you can at least list out the SC and Q, and then they have some examples in there. So just leaving breadcrumbs for people if they want to explore.

Underrated Prompting Patterns and ELI5

Vibhu [00:30:31]: Before we move on, I wanna ask you, any other underrated tips, ways people could get a lot of value from Claude Code that they’re not using?

Thariq Shihipar [00:30:41]: Yeah, I think a lot of them are in the, this unknowns, like, doc. Like, I give a bunch of example prompts, like, using it for brainstorming, using it to quiz you after. we added this, like, explain it like I’m five skill which is a very short prompt. And it doesn’t even say explain it like I’m five. It’s like the key word of this prompt is big pictures, few words. like, that’s like the main thing. And it is shockingly good? Like, you, like, I think I tweeted about this and it’s like /eli5, and, like, you can install it as a plug-in. But yeah, it’s, like, way better at just cutting through the BS and being like, yeah, exactly right here. So the diagrams are, like, quite clear. I think one of the things that is true with artifacts is, like, they put too much text in and people are not reading the artifacts? And so, like, this simplifies it a lot more. And, yeah, this came out of, like, just people at Anthropic, like, going through very complicated incidents and being like, “What is happening?”? So, this one I think is great, yeah.

Swyx [00:31:47]: My version of this is the, it’s like test your understanding. Give you a few choices and then, like, if you get it wrong, you have a mismatch between what you think is happening versus what’s happening.

Thariq Shihipar [00:31:58]: Yeah. I think this is one of those things that everyone loves talking about, and then very few people really do. Like, I think

Swyx [00:32:05]: Really helpful.

Thariq Shihipar [00:32:07]: Yeah. But most people just don’t want to get quizzed about something? Unfortunately, I think this is one of the, like, things that we need to, like.

Swyx [00:32:16]: What’s the opposite of ask you the question or ask you the question before the thing?

Thariq Shihipar [00:32:19]: Yeah.

Swyx [00:32:19]: This is after the thing.

Thariq Shihipar [00:32:20]: Exactly. Yeah.

Vibhu [00:32:21]: It’s a good way to stay grounded of, like, do you even know what you’re doing, right? The worst case is when people send you slop and they haven’t understood what they’re asking for or what the output is, and it’s like, “Dude, I don’t wanna read this. Do you even know what it is?” So, you make it a rule for yourself that before you send stuff, you should at least know what’s implemented.

Claude Mods: Customizing the Harness

Thariq Shihipar [00:32:41]: Yes, but so you could make this a mod and you could build your own mod to, like, make sure you test it. So yeah, you can do that.

Swyx [00:32:49]: All right. Let’s get right into it. What is Claude Mod, and what is this diagram showing?

Thariq Shihipar [00:32:54]: Yeah. Okay, so Claude Mods is you can customize the entire Claude Code harness, and we’re going to. If you have requests, we will, like, let you, like, please let us know. We’ll add more and more. This works for CLI, it works for desktop. maybe it will work for Claude Tag in the future. I don’t know. Like, we’re trying to make this very extensible. You can see this reference sheet. I don’t want people to get overwhelmed by it? At a high level, you can customize both the execution of the harness, and the UI of the harness. And so, like, you say on that Tetris example from Boris, that’s like customizing the UI, right? Like showing, like, Tetris in the game.

Thariq Shihipar [00:33:35]: But, like, let’s say that you wanted to do this thing where you had. you tested your assumptions or, like, tested your understanding after every project, right? What you would do is you would ask Claude to make this plug-in. It would spin a classifier after every prompt. And so, like, at the end of each turn, you would spin off a sub-agent or, like, a forked agent. A forked agent is, like, maintains the prompt cache, right? So it’s like a, like one of those unintuitive things where you can fork and do, like, a little request, and it’ll be very cheap because the entire prompt cache is, like, done. And so you can be like, “Has this task been completed?” like

Swyx [00:34:18]: This is how you do BTW and all those.

Thariq Shihipar [00:34:20]: Yeah. The underlying forked agent, yes. But so you can, in the f-fork sub-agent, you can say, like, “Has this task been completed? If so, return true.” And then in your hook, or in your, like, plug-in mod, or sorry, like, in the sub-agent probably, you would say, like, “If true, give me a quiz.” give me questions and answers, and then, like, in a JSON format, and then you’d parse it, and then you display above the prompt input, this list of questions, right? And so this is something that’s, like, slightly token-intensive because, like, you have to do it after every end of the assistant turn. But it’s, like, a lightweight classification, and then you can, like, get this quiz, and then you’ll see, like, Claude will always do it for you. You don’t need to remember to do it. There are lots of these, like, tips that we’ve talked about, right, where it’s like, oh, implementation notes. You can also add a tool for implementation notes now. And so, like, this tool that I’m adding is, like, register, like, I think assumption is what I’m calling it, but, like, maybe I’ll change it around. And this is a mod. And so, like, you give it a register assumption tool, and then it will keep a list. It’ll. Every time it does it’ll keep a, like, add to the list, and then at the end it will display those assumptions? Another mod I’m working on is a model router. And so, like, internal, like, Claude model routing, right? So it’s. This is, I want to say the reason we don’t do model routing by default is, like, it’s a hard problem? And like

Forked Agents, Assumption Tracking, and Model Routing

Swyx [00:35:51]: You will get it wrong.

Thariq Shihipar [00:35:52]: Yeah, you, like, yeah, you will, like, accidentally use, like, Fable for a hard problem or Sonnet for

Swyx [00:35:57]: Yeah, if you have auto approve, but you don’t have auto mode.

Thariq Shihipar [00:36:01]: Well, you will have auto. Like, you don’t have, like, auto routing or something.

Vibhu [00:36:04]: You don’t have auto mode for model picker.

Thariq Shihipar [00:36:06]: Yeah, exactly. So

Vibhu [00:36:07]: I’m getting the rough question of, like, how much do you open this up and how much do people have to think about this? Like, when you talk about prompt caching and building a router, it seems like you could easily build a mod that routes per query, and I’m just killing my plan very fast, right? I guess my question is more so, like, what is, like, a product talk like this look like, right? Who is it for? Is it for power users? Is it everyone should be able to go through

Swyx [00:36:33]: Oh, definitely power users, right?

Thariq Shihipar [00:36:35]: Yeah, I think it is power users, but, like, the nature of Claude Code is that so many people are power users? Because it’s easy to share things, like you can. Like, one person can make a good model router thing that doesn’t break prompt cache all the time, and then you can, like, compose them. Another cool thing about the plug-ins is that they can hook into and compose with each other. And so I have, like, a mod that will, like, create a mode selector at the top, and any plug-ins can register to be a mode. And so, like, the auto router can be a mode, right? Or, like, you can have a mode that’s, like, artifact mode, where it’s like it primarily talks to you in artifacts. like, you can toggle between plan mode? And so, like, you can create more and more of these modes. But the ability to create modes is in it itself a mod? And so there’s a lot of richness here, but we do want to make it fairly easy. We want to be-- make it so that you can just, like, install someone else’s. You can ta-- you can chat with Claude and, we’ll, like, make sure that it understands the nuances of things like prompt caching and stuff, so it can, like, warn you. This is, like, not extremely complicated behavior for Claude, I think, but we should have just a good skill on how to make mods. and yeah, we’ll see how we go. But I do think that this is, like, a preview of, like, mutable software, and, like, how, like, generative software, just like you can customize safely. If enabled, you could customize any piece of software. And I think that more and more apps ideally do something like this?

Power Users, Modes, and Mutable Software

Swyx [00:38:13]: And by the way, you, we have, you have another cool tweet about how, there’s the infinite money button, which is like make your SaaS, consumable by agents. I think mutable software is interesting and, other people have also tried to do it. I think the hurdle comes when you can do everything, then people, users get, tend to get confused. So usually the stuff that works is just like one opinionated flow. This is in the side of less opinionation. It’s just like, well, more power to power users. And I think probably unlocked by AI, where, like, you can just prompt for whatever the thing is.

Thariq Shihipar [00:38:47]: Yeah, or there can be a skill that gives the opinions?

Mods vs. Hooks vs. Artifacts

Swyx [00:38:50]: Yeah.

Thariq Shihipar [00:38:50]: And then, yeah.

Swyx [00:38:51]: So knowing a little bit about, like, TypeScript and build systems and all these things, the closest-- I’m very curious that the team who worked on this, if, I don’t know how close you were to them, if they drew any inspiration from build systems like Babel, Webpack, all these, like, old school things. Because it sounds very similar, like the plug-in ecosystem of those things where they can compose with each other.

Thariq Shihipar [00:39:11]: Yeah, I’m not deep in the technical details, but I do know it was a collaboration with someone on the Bun team and someone on the Claude Code team.

Swyx [00:39:17]: Yeah, it’s a build system mecca.

Thariq Shihipar [00:39:19]: Yeah. Exactly. It’s, it’s very exciting. But yeah, like, agents can just do this very complicated like, extensibility into your software now. And so, yeah, like, another reason to, like. If you run a startup, like, you can just prompt Claude and be like, “Hey, like, could we make an extension system? Like, what would that look like?”?

Swyx [00:39:37]: Yeah.

Swyx [00:39:38]: And I just really wonder, like, you had hooks in the past and plug-ins, all these things. So what specifically will mods be able to do that those things could not do?

Thariq Shihipar [00:39:47]: Internally, we were originally calling this function hooks. And so, like, that’s, like, gives you a little bit of an idea where, like, hooks register a, like an event to happen and then, like, a script to call. And this inside of the, like, TypeScript runtime is running things. And so, like, you get some benefits of just, like, it has a bunch of things in the Scope with, like, for example, like how many turns is in this conversation, right? Like, how many tokens have been used? Like, et cetera. Like, what are the messages? Things like that. So it has a bunch of messages that can be used. And then it’s just, like, a lot more hooks. So we have, like, or a lot of, lot more, like, things you can register on. And then you can do because of the. because it’s all happening in process, you can, spawn sub-agents, with four contests and contexts and stuff. And, like, that will return. You can parse the results of those. You can use structured output to like, return them. and then you can modify the UI, which you can never do in hooks. So, yeah.

Swyx [00:40:50]: Yeah. Yeah. So modify UI, this is why you showed the Tetris example. Does it also ex-extend to artifacts? I assume it does.

Thariq Shihipar [00:40:57]: You-- Like, artifacts are like a different way of customizing it. like, you can definitely. One of the mods I’m working on is, like, this dashboard mod, which will, like, prompt Claude to maintain a dashboard, that’s an artifact. But they’re like, slightly orthogonal, or not orthogonal. They compose with each other in different ways. Like, mods are, like, a little bit more, like, in your Claude Code harness, changing the agent loop? And, like, the UI is, like, an added benefit. and then artifacts are just like you want to, see things at a high level, very inter- highly interactive. like, the affordances can be a lot bigger than, like a TUI or even in our desktop.

Next Steps, Supervisors, and Persistent Guidance

Vibhu [00:41:40]: I’m guessing you’ll have a good blog post on the differences, because right now you can also, make a loop that outputs to an artifact that’s an interactive dashboard, but you can also do it with a mod. There’s just some thinking about making a hacking on a harness when we don’t know much about the harness, right?

Thariq Shihipar [00:42:00]: Well, something I’m excited about with mods is, like, there’s so much things with Claude Code that you just have to remember? You’re like, “Oh, like, let me do this, and then let me call the dashboard skill that does the loop,” and things like that. And, or like, “Let me test my assumptions afterwards.” And I think, like, if you do all of these things using these little classifiers and stuff, and you’re like, “These are the things I care about. This is what I want to do,” you can, like. You don’t have to remember as much. One more, like, mod I’m working on is a next steps mod that

Swyx [00:42:28]: I have-- I was gonna say, I have a next step skill. I always run next steps.

Thariq Shihipar [00:42:32]: And does it have access to your skills? Like, this is one of those things where I’m like.

Swyx [00:42:37]: I think so.

Thariq Shihipar [00:42:38]: Okay. Yeah, probably

Vibhu [00:42:39]: Do skills need specific access to

Thariq Shihipar [00:42:41]: Well, I think there’s

Swyx [00:42:41]: Don’t they always have

Thariq Shihipar [00:42:42]: I think there’s, like, specific prompting, I guess, to, like, know your skills. Like I think Claude forgets them sometimes throughout, like, the thing. But anyways, the idea of, like, yeah, next steps that also are like, “Oh, hey, this has happened. Use the explain skill to explain to you what happened because this seems, like, quite complex,”? Or, like, yeah, “Use your unknown skill. It looks like you are, like, asking the model to, like, iterate on these small changes. It seems like you could prompt better.” like, “What if you did this?” Right? So, I think, yeah, like spending more compute there. Yeah.

Swyx [00:43:20]: And it should always come out as multiple choice. we have, I have

Vibhu [00:43:23]: We have his skill.

Swyx [00:43:24]: My next step skill is like this.

Thariq Shihipar [00:43:26]: Okay, perfect. Yeah.

Swyx [00:43:27]: You can steal it.

Thariq Shihipar [00:43:28]: Yeah.

Swyx [00:43:29]: Like, but like, for me, it’s all-- I think models really always need to be reminded, what are you trying to do here?

Thariq Shihipar [00:43:35]: Yeah.

Swyx [00:43:35]: Look at the whole transcript and go like, oh, was this original goal? Did your solution solve it? Were you lazy? If you’re lazy, maybe there’s a reason. Maybe you needed approval from me. Maybe you needed, there’s two things you wanna suggest. So it’s, it’s a little bit like the modification of the ask user question or interview me skill. so it’s next steps.

Thariq Shihipar [00:43:55]: Yeah, exactly. And again, the benefit of doing it with mods is you can do it as a fork sub-agent, and so it doesn’t remain in the context afterwards. So you have this, like, idea of like, okay, the model is doing its execution and you have this almost like supervisor, like, that is like making sure that you can do like the next steps well. So yeah.

Swyx [00:44:15]: Yes. I do have two panels and like I often try to have a supervisor thing, keep the high-level context and then the implementation

Thariq Shihipar [00:44:21]: Yeah

Swyx [00:44:22]: Detail in another agent.

Vibhu [00:44:23]: I feel like a lot of this abstracts away as models change? The, like, half an hour ago you said bitter lesson of harness engineering

The Bitter Lesson of Harness Engineering

Thariq Shihipar [00:44:31]: Yeah

Vibhu [00:44:31]: And we’re on the other extreme right now, I feel.

Swyx [00:44:33]: Well, so yeah, exactly. If everything’s customizable, what is Claude Code, right?

Thariq Shihipar [00:44:37]: Yeah.

Swyx [00:44:37]: And which I talked to you about last night.

Thariq Shihipar [00:44:40]: Yeah, I think that this is. I think the bitter lesson is unintuitive? In terms of like. Also, like we’re misusing a little bit of the bitter lesson here where it’s like, it’s more about like scaling and compute and stuff. But like, I think there is something where it’s just like. I think I use it as an approximation here to say that harnesses go out of date very quickly? And like how, but how they change is unintuitive? And so like the big obvious example is like from chat to like agents where you had to give them entirely new tools, right? But like, I think this new version of like, oh, it can modify its own harness, right? This is like, an own harness loop is like a way of using its capabilities, right? Or like it can build an artifact. And like, I think the way I think about it is like the models have more and more intelligence, and they’re like so much more intelligent now than like the average software engineering task. Like, you look at the like terminal bench ones and they’re like solve like the Jacobian conjecture. Not really, but like, it’s like they’re, they’re quite complex. Like, I would not have been able to do this really as a software engineer.

Swyx [00:45:42]: And you said TB4 or TB2?

Thariq Shihipar [00:45:43]: TB3. TB3.

Swyx [00:45:44]: TB3.

Thariq Shihipar [00:45:44]: Yeah. They’re quite complex, but the goal is still to deliver user value, right? And like you said, there’s like this infinite space of things to do. And so the ways like you spend compute are to keep the user in the loop and make sure that like you’re getting to the right decision in the end of the day and like the right output. And artifacts and mods are this way of like spending that intelligence. and I think that’s like, yeah, the next step. And so, yeah, I think Claude Code is like, has the core things of agent loop which are, have gotten more complicated. It’s like, it needs a sandbox to operate safely. It needs auto mode to like make sure like the permissions

Vibhu [00:46:21]: Approvals.

Thariq Shihipar [00:46:21]: Yeah, approvals. it needs computer use and MCPs and like all of these like ways of accessing your data, and it needs web search and web fetch. And like, so the-- as the models can do more and more, the core harness has to be like quite complex and very secure. But then like how you interact with it can change quite a lot.

Vibhu [00:46:42]: What other harness engineering best practices have you, from the Claude Code team itself? I feel like, there was a phase of plan mode, which is not as used. We now have auto mode. at a point you cut the majority of the system prompt, you got rid of examples. What other best practices are there for harness engineering?

Core Harness Primitives and Managed Agents

Thariq Shihipar [00:47:02]: I think there is like a forking path where at some point, eventually, yes, the model will just be able to like vibe code the exact version of Claude Code, even describing all this complexity that I’ve talked about, right? Like auto mode and computer use and stuff. Eventually, the models will just be able to do that in one shot. But I think they can one shot simpler harnesses? And so like, I think some people. Sometimes you don’t need this full, like if you don’t need computer use or like all this like more complicated stuff. I think before we, you had to use things like the agent SDK, which was like Claude Code wrapped, in order to like. And I would, like suggest people do that because there was so much complexity into building a harness. And now as that’s got more abstracted, we have like, Claude managed agents, which lets you have that complexity, but still like, right, like a very bare bones like harness that’s scoped to your task. Yeah, I think there’s like this barbell effect where like for like very complex, for like …

From the earliest days of open-weight models to becoming the neutral routing layer for more than 10 million developers, OpenRouter is one of the clearest bets that the future of AI will be multi-model. In this episode, OpenRouter co-founder & CEO Alex Atallah, with AMP’s Anjney Midha returning with swyx to unpack how OpenRouter emerged from the first wave of Llama, Alpaca, Mistral, and Midjourney, why model diversity mattered before it was consensus, and how a company dismissed as “just a wrapper” became critical infrastructure for the AI ecosystem.

We go deep on the product and distribution lessons behind OpenRouter: why model labs can spend billions training a checkpoint and still struggle to get it into developers’ hands, how Mistral helped prove the value of a competitive inference marketplace, why OpenRouter chose focus over expanding into fine-tuning, memory, and other adjacent products, and how its rankings became a real-time map of how AI usage was changing. Alex also explains OpenRouter’s early experiments with model fusion, why they deleted the first version and brought it back years later, and how the platform grew to more than 10 trillion tokens per day.

Finally, Anjney explains why Stripe and OpenRouter fit together, why token fraud may become one of the defining security problems of the AI economy, and why the next wave of fraud won’t just come from humans but from autonomous agents attacking increasingly valuable token flows.

We discuss:

* Why OpenRouter bet early that no single AI model would win everything

* Alpaca, Llama, and open models becoming impossible to ignore

* Why Discord’s early AI deployments exposed the limitations of closed models

* Why model labs can spend billions on training and still fail at distribution

* How OpenRouter became a neutral distribution layer for model developers

* Why VCs dismissed OpenRouter as “just a marketplace” or “just a wrapper”

* The Mistral price war and the first real proof of an inference marketplace

* How Midjourney scaled through Discord and what it taught the AI ecosystem

* Why crypto infrastructure became a dress rehearsal for generative AI

* OpenRouter vs. LM Arena and why their missions are fundamentally different

* Why focus became one of OpenRouter’s biggest strategic advantages

* Anthropic’s early focus on AI pair programming and coding

* The OpenRouter products that were prototyped but never launched

* MOM, OpenRouter’s early Mixture of Models experiment

* Why model fusion failed in 2024 — and why it works much better now

* How OpenRouter’s leaderboard became a live map of the AI industry

* OpenClaw, auto-routing, and agents reshaping AI usage

* How OpenRouter reached 10+ trillion tokens per day

* Why inference gateways are increasingly becoming targets for fraud

* Why Stripe’s fraud infrastructure is strategically important to OpenRouter

* The coming rise of agentic fraud and attacks on the token economy

* What changes and what stays the same as OpenRouter joins Stripe

Alex Atallah

* LinkedIn: https://www.linkedin.com/in/alexatallah/

* X: https://x.com/alexatallah

* Website: https://alexatallah.com/

Anjney Midha

* LinkedIn: https://www.linkedin.com/in/anjney/

* X: https://x.com/AnjneyMidha

* AMP: https://www.amppublic.com/

Timestamps

00:00:00 Introduction

00:02:12 Alpaca, Llama, and the Multi-Model Bet

00:06:04 Discord, Open Models, and OpenRouter’s Origins

00:14:28 Why “One Model Wins” Was the Wrong Bet

00:17:27 Why Model Labs Struggle With Distribution

00:23:04 “Just a Wrapper”: Why VCs Misunderstood OpenRouter

00:27:58 Bootstrapping OpenRouter Through Community

00:36:16 Crypto, Midjourney, and the Early Generative AI Ecosystem

00:43:38 Mistral and the Birth of the Inference Marketplace

00:47:10 OpenRouter vs. LM Arena

00:52:08 Focus, Anthropic, and Roads Not Taken

00:59:34 Mixture of Models and Model Fusion

01:02:44 Sonnet, OpenClaw, and OpenRouter’s Explosive Growth

01:09:03 Why Stripe Acquired OpenRouter

01:12:45 Fraud and the Emerging Token Economy

01:17:47 The Coming Wave of Agentic Fraud

01:19:07 What’s Next for OpenRouter at Stripe

Transcript

Introduction: OpenRouter, Marketplaces, and Pub-Sub as a Product Principle

Swyx [00:00:00]: Okay, we are here in Anja’s house, which is where all big startups in San Francisco start.

Anjney Midha [00:00:08]: Howdy.

Swyx [00:00:08]: And, congrats on Cursor, Mistral. I don’- God knows what else. You got so much stuff going on.

Anjney Midha [00:00:17]: There’s, there’s a lot going on. Well, OpenRouter is probably the - has been the most, I would say, like, one I’m excited about recently.

Swyx [00:00:24]: Yeah. And we have Alex, first time on the pod, but,

Anjney Midha [00:00:27]: Thanks for having me.

Swyx [00:00:27]: You’ve been in the IE a few times. I appreciate every time you’ve shown up, for the community. Congrats. I just, like, what a journey. When I was looking back at your past posts, one of the earliest principles that I saw you write as a product person is sub as a product principle. And I wanted - you to maybe explain how you think about what should exist in the world.

Anjney Midha [00:00:49]: Yeah. The sub piece, which was early 2023, I didn’t think about it until we talked like 10 minutes ago, is about how there is like a way of thinking about products as an intersection between subscribing to data and publishing data. And marketplaces are an easy example of this. You have suppliers that are publishing some product to a SKU. And the SKU is like a sub topic that a consumer is subscribing to and just going to, like, consume whenever they want. And humans consume in a very, like, discreet, ad hoc way. It’s not very scalable. all their attention is on the topic when they’re buying the thing, and their attention is nowhere else when that happens. agents and consumers of inference don’t act like that. They’re consuming continuously, and they’re changing the SKUs that they consume from all the time. So OpenRouter is like a blend between a normal API experience and a marketplace where we create model slug. We have the auto router. We have all kinds of, like, product SKUs that you can subscribe to. And then you can, like, continuously add, like, derive value and make decisions based on those consumers.

Alpaca, Llama, and the Multi-Model Bet

Swyx [00:02:11]: Yeah. This is something that was more consensus now, but not consensus when you guys started, which was that there is such a demand for swapping models and changing things out and, that people would not use the native SDKs. I guess, for each of you, what was your realization moment that this would be it? I, - You’ve, you’ve given a talk at EIE about Alpaca as,

Anjney Midha [00:02:33]: Yeah.

Swyx [00:02:33]: One of your inspiring moments.

Anjney Midha [00:02:35]: Alpaca, I can, like, rehash the Alpaca moment for a sec. Like, the very beginning, at the end of 2022, OpenAI was the only game in town. There was, like, OpenAI, Cohere,

Swyx [00:02:47]: Yes.

Anjney Midha [00:02:48]: And then a smattering of, like, early attempts at open weight models.

Swyx [00:02:54]: Yeah.

Anjney Midha [00:02:54]: When Llama came out in January of 2023, it was like, “Wow, really exciting. This is really big.” It outperforms 3 on, one or two benchmarks. but you can’t chat with it. It wasn’t like - It wasn’t an engaging model, but it seemed like someone just needed to fix a couple things and do some RLHF on it to get it all the way there. And Alpaca was the first model that I saw that did that. It only took $600 to do. A team at Stanford generated a bunch of synthetic data, tuned Llama, and made Alpaca, billion parameter model. Or was - Maybe it was thirteen billion parameters. And it was so good. Like, I was just, like, on an airplane using it. I, - in many cases, I, like, you could not discern a ChatGPT versus an Alpaca result. And I figured if it was this easy to make a model, one, we have a whole new way of monetizing data for the first time. you can just, like, take really valuable data and turn it into a service in $600. and that cost will probably go down over time.

Swyx [00:04:03]: When you - So sorry. when you say monetizing your data as, what eventually will become an MCP endpoint or as a training data for a model?

Anjney Midha [00:04:12]: Yeah, training data for a model.

Swyx [00:04:13]: Awesome.

Anjney Midha [00:04:13]: Like, an abstract way of saying like, “Hey, I have this data.”

Swyx [00:04:15]: Compress it into a model.

Anjney Midha [00:04:16]: Like, it makes sense for me in my product, but, like, I could repackage it in the form of a model and sell it. And so it’s just a whole new business model for the economy. It also, of course, provides, like, a way of following what Frontier Labs are doing, but in a way that, like, a single developer or a small team of developers can roll on their own. And so - Whenever you have an example of that, like a breakout app that’s doing really well, and then some framework for imitating it with - in your own flavor, you have an immediate ecosystem of, like an immediate ecosystem, like, should arise because there’s just a huge gap between the, like, decisions that the single company is making and all of the variations in those decisions that, like, a wider ecosystem can create themselves. And so then, you need a marketplace to, like, discover all of those, services and all of those products. There wasn’t any place on the internet that, like, was like a home base for LLMs in terms of seeing how much they were being used and seeing who was using them and why.

Swyx [00:05:29]: The closest would be Hugging Face.

Anjney Midha [00:05:30]: Hugging Face was the closest at the time, yeah.

Swyx [00:05:31]: They just started Hugging, like, a few years ago before that.

Anjney Midha [00:05:34]: Yeah, and Hugging Face also didn’t have the closed-source models.

Swyx [00:05:37]: Yeah.

Anjney Midha [00:05:38]: And they didn’- you couldn’t use the models at the time. and there wasn’t data about who was using them. There were, like, a bunch of differences between OpenRouter and Hugging Face, and those differences felt really critical to me, especially when I was just trying to learn about LLMs and, like, why people are choosing, like, Different little ones that are emerging over time.

Discord, Open Models, and the Origins of OpenRouter

Swyx [00:06:03]: Got it. And then, Ansh, no stranger to wanting more model diversity, at the time, you’re a couple of years into your Anthropic journey, which we covered in the previous podcast as well. What was your introduction to Alex?

Alex Atallah [00:06:16]: Well, the introduction was, I think, thirteen years before that.

Swyx [00:06:20]: Oh.

Alex Atallah [00:06:20]: But the OpenRouter handshake happened right over there, if you remember.

Anjney Midha [00:06:23]: Yeah.

Alex Atallah [00:06:24]: Which - So Alex and I, met, I believe as sophomores now, if I remember at the Stanford Review,

Anjney Midha [00:06:32]: That’s right

Alex Atallah [00:06:32]: Meeting for the first time.

Anjney Midha [00:06:33]: I think so, yeah.

Alex Atallah [00:06:35]: Yeah.

Anjney Midha [00:06:35]: Yeah.

Alex Atallah [00:06:35]: So Stanford Review was the libertarian newspaper on campus at Stanford that Peter Thiel started back in the day. And, whatever-- for whatever reason, I, Alex and I both showed up to one of the meetings, and I remember, the editor-chief was a mutual friend of ours. Lisa was really a really great editor-chief, where, part of an editor-chief’s job is to assign responsibilities to people and make sure the work gets done. and I, I may be misremembering the details, but I remember wanting to. It was surprising to me that at the time there was no dedicated technology section in the newspaper.

Alex Atallah [00:07:11]: You

Swyx [00:07:13]: Because it’s political, right?

Alex Atallah [00:07:14]: It is primarily

Swyx [00:07:14]: Like, it’s talking

Alex Atallah [00:07:15]: It originally started as like a

Anjney Midha [00:07:16]: Yes.

Swyx [00:07:17]: Yeah, states and all those things.

Alex Atallah [00:07:17]: Correct.

Swyx [00:07:18]: Yeah.

Alex Atallah [00:07:18]: But it, - To take us back in time, you may remember this, but, there was this technology, legislation that was being debated called, the Net Neutrality Act. And net neutrality is, like, inherently this political concept, right? It’s, it’s about the regulation of - internet broadband access. And so there was a community of us who were technologists, but also debating the politics of the technology. And I thought the Review would be a great place - to, like, write about that. And I was working on, I think, a net neutrality article, and I remember proposing, “Well, maybe we should start a technology section.” And Alex was one of the only people who said, “Yes, that would be cool.” And said. I forget whether we ended up writing stuff together, but - that’s when we first met,

Alex Atallah [00:08:03]: Was 2011 or twelve. I forget which year it was. It was one of those.

Anjney Midha [00:08:09]: Yeah.

Alex Atallah [00:08:09]: It was at Old Union, if I remember correctly.

Alex Atallah [00:08:11]: That’s where we used to meet. But, along the way, Alex and I have had a chance to, To hang out often. And probably the time when we had the most professional overlap was when I was running the platform at Discord, and it had become this explosive platform for crypto

Swyx [00:08:32]: Yeah

Alex Atallah [00:08:32]: And NFTs in the middle of the pandemic.

Swyx [00:08:35]: Which also, by the way, you were in charge of safety and security as well, right?

Alex Atallah [00:08:38]: I was the head of platform, which meant all of the crypto - the DAO and NFT launch security debugging fell on

Swyx [00:08:45]: And their phishing and.

Alex Atallah [00:08:47]: The phishing, the social engineering attacks, the katana DDoS that we were getting hit by. but it’s around the time I first started teaching security at scale at Stanford, CS 153. And Alex was on the, - at OpenSea at the time, and I was trying to figure out how we could defend against all these attacks that we were. Like, and at peak, I forget, if you remember how much NFT volume was running through

Swyx [00:09:10]: Discord

Alex Atallah [00:09:10]: Discord, but it was, like, a meaningful amount of, like, it was, like, several billion dollars in NFT volume of GMV, so to speak, were running through the platform, and it was all coming from OpenSea. It was these, like, buy, sell,

Swyx [00:09:20]: The

Alex Atallah [00:09:21]: Servers

Swyx [00:09:21]: The D in DAO is Discord.

Alex Atallah [00:09:25]: Yes. And so that’s when I think we had hung out professionally. But a year after that, OpenAI gave Discord early access to GPT. Sorry, three. No, it was five. Yeah, five, which is the RL version of three. And that’s around the time we made a Discord bot with, OpenAI for internal deployment, and that’s when I realized we would need. Like, since I was part of the deployment team.

Anjney Midha [00:09:50]: What was the use case?

Alex Atallah [00:09:51]: There were two that were. And there’s, there’s a post now called “Discord is Your Place for AI with Friends” that somebody sent me recently that I wrote, and published in twenty-three. But There were two use cases. One was Clyde, which was the - like, a party friend inside of Discord that could help you set up your Discord server and talk to you about onboarding and get your friends to hang out more. and then there was content moderation. And one of the realizations we had with content moderation was - it would refuse to moderate. Like, it would just refuse our prompts because the The training was. We were very early in the training era, and it would just. Our prompts would trigger it, its, like, guardrails. And we told OpenAI, “Hey, guys, we need access to the weights because if we’re gonna be doing content moderation at scale, we had 250 million monthly active users, we need more reliability that the model will do what we need it to.” And they said, “Well, sorry, guys, that’s not how this works. We’re a closed-source company.” And so that was my first realization that we needed open models, and the enterprises would need more control over capabilities, and then ultimately would need some control plane or management system to orchestrate these open models. But there weren’t no good - there were no good open alternatives until maybe

Alex Atallah [00:11:10]: Six months later when Llama came out. And six months after that, I led the series A into Mistral, which was started by Guillaume and the Llama team. And - That, - Around that time is when I remember hearing about Alex launching OpenRouter and going, “These worlds are gonna collide, and I don’t know when it’ll make sense to team up.” But Alex was so early and could see. I think he was totally right about this ecosystem starting with Llama that then needed, like, a, an easy layer to manage for, especially for. I was approaching it from the enterprise perspective because I had been that, like, the. As the VP of platform at Discord, it was my job to ensure that when we deployed models to, like, 250 million users, they did what we wanted them to. And that was very hard, because if you outsourced it to the labs and they controlled the guardrails and their guardrails are their safety policies. Forbid the model from responding to your prompts. That was quite catastrophic.

Swyx [00:12:05]: Yeah. But what, a moderation is the thing that they want to support. And obviously, beyond that, they would - OpenAI would work with you, presumably to give you a moderation endpoint, which they offer for free.

Alex Atallah [00:12:16]: It was an interesting use case, that - So they did give us a moderation endpoint. However, as you guys know, every Discord server is like a mini deployment of itself. And so the use case was instead of having human moderators that have to interpret the norms of the community, you just give the, - Often, like every, subreddit, Discord servers, public ones have their own rules that the user, the users create.

Swyx [00:12:41]: Oh, yeah. We run the LinkedIn Discord in. Yeah.

Alex Atallah [00:12:43]: And then humans used to read those norms and then enforce it every day manually, like observing each message in these communities. And these communities have like millions of users. So we had a 5,000+ person team globally in the, on the Discord content moderation team. These are outsourced contractors who had a really tough job. And so the idea was instead, if you could give the norms of that server To the LLM, then the LLM would do custom moderation for that server. It’s almost like a, like context moderation for that server. And many of those servers’ norms just violated OpenAI’s rules. And so - It was like we had our own custom eval. So each server had its own custom eval. But Discord-- at the time, OpenAI’s evals, we were all so

Alex Atallah [00:13:28]: Primitive in our thinking about how to deploy these LLMs that often the training prompts were super handed. It said, “Oh, anything about Harry Potter, anything that has trademarked content, don’- refuse.” And if it was a fan - Harry Potter fan community, this is a real use case, that had content moderation, the LLM would just refuse.

Swyx [00:13:48]: Yeah.

Alex Atallah [00:13:49]: And that was just not precise enough.

Anjney Midha [00:13:52]: Another one that we heard was like if someone was trying to write like a detective story, and there’s one chapter with a lot of violence, like maybe someone

Alex Atallah [00:14:01]: Right

Anjney Midha [00:14:01]: Like kills someone, the LLMs would just refuse to, like, help with that part of the story.

Alex Atallah [00:14:07]: Yeah.

Anjney Midha [00:14:07]: And then - like, we used to be like, okay, this is not like structurally inherent to LLMs. There must be, like, some choice out there so that I can, like, switch to another model, when I’m getting, like, a refusal or a bad result from the main one that I have. And that, like, tension also drove me for a marketplace.

Why “One Model Wins” Was the Wrong Bet

Swyx [00:14:28]: Yeah. I think that is well accepted now. What was it like back then when you were raising or, starting this? did people get it? what was the, some of the struggles? I like getting stories out of him about how other VCs don’t get it. So like anything you wanna, talk about, now - Let’s, let’s call it, that the early journey of OpenRouter is done, right? You can obviously talk about some of the early days stuff.

Anjney Midha [00:14:54]: Well, I was gonna say that, like, the biggest objection we got is big model win, which is - all of the

Swyx [00:15:03]: Scaling laws.

Anjney Midha [00:15:04]: Huh?

Swyx [00:15:04]: Scaling laws.

Anjney Midha [00:15:05]: Yeah, scaling laws, and natural network effects are just gonna accrue to one company, which will be - It’ll be a Google-style monopoly, just like how Google won the search market, by a large margin, and you’ll just be fighting for scraps at the end. That was probably the biggest objection we got. it is interesting that Google won the search engine race with such a huge margin. I think, like, had there been more interesting benchmarks or had, like, search engines been, - had people, like, seen them a little bit more like LLMs where they’re services that you can build companies on top of, that might not have been the case. but LLMs don’t merely have a user interface. They’re also, like, ways of building entirely new businesses. And, a Google-level monopoly would be like the Dutch East India Company times, quadrillion in magnitude because the whole economy ends up, like, depending on the one monopoly as well. So it didn’t seem like would be a really crazy outcome if that happened. And it’s also less likely because the economics of, like, creating good competitors are much, like, much more decentralizable.

Alex Atallah [00:16:25]: Everything Alex said is true, And I came at it from a completely different perspective, which

Swyx [00:16:31]: Yes, this is why we’re here.

Alex Atallah [00:16:32]: The scaling laws were never - In my mind, were always a feature, not a bug for why OpenRouter would be very valuable. Because, I was one of the first investors in Anthropic, and it was obvious to me that other researchers in our friends - I went to grad school for machine learning, and I just had a lot of friends in the ML community who it was very obvious to us that the bitter lesson holds. And so I was like, “Oh, fantastic. Now we have at least two proof points that compute scaling works.” It was OpenAI and Anthropic. and by the time I think we decided to team up on OpenRouter, I had already invested in Mistral and Black Forest Labs and Luma. So there was multiple model companies and teams that I was, working with.

Why Model Labs Struggle With Distribution

Swyx [00:17:14]: But you did other modalities, whereas this is literally

Alex Atallah [00:17:16]: Across different modalities, yes

Swyx [00:17:17]: Text.

Alex Atallah [00:17:18]: Exactly. And it was so obvious to me that an ecosystem of different kinds of models were being created, and that this whole narrative of, like, Only one company will dominate like Google was, well, like maybe true, but one, I don’t believe that. But two, there was so much extraordinary innovation happening across several different research teams. But the shared problem I was noticing across all of them was often, the research teams were fantastic at figuring out how to reason about new capabilities. They think in terms of capabilities, but never - like, are not developer mindset-oriented. Like, what happens after the training is done and the checkpoint comes out? Like, you’d be shocked how, like, similar the early training teams at OpenAI, sorry, Anthropic, BFL, Mistral, were in their, like, default approach to. Taking their research out of the, lab and scaling their impact, which is often, oh, the checkpoint is done, put it out as an API, done, and then there’d be crickets. in the case of Claude, the first Claude checkpoint was done a year before they released it internally. And then ChatGPT came out, and we decided, okay, yes, it’s a good idea to release a Claude version externally.

Alex Atallah [00:18:34]: And they had no plan, like no plan for how to get developers to try it out. And so if you go to the Claude one blog post, you’ll notice there are, like, three developer examples for users of the API, and one is a Discord bot, and the second is Vivian, my wife’s startup called Juny Learning, ‘- And then there was, like, Notion, because these were all friends of, like, the Anthropic Because that’s how - like, last minute the planning was around, hey, once the model’s done training, how do you get it out to the world? There was no distribution platform that understood what developers needed, all the key management, provisioning, like, simple, like, endpoint management, versioning control. Like, all these things that the scientists and researchers go, “ that’s plumbing. I don’t really think about it.”

Swyx [00:19:15]: Implementation detail.

Alex Atallah [00:19:16]: Right. And instead, Alex came at it from that perspective. And so, it was so obvious to me that, like, every single lab I was funding would spend - like, literally sometimes billions of dollars into training, and then a checkpoint would be done, and there’d be crickets, like, during early access because they’re like, “Oh, that’s right.”

Alex Atallah [00:19:35]: It’s hard to use a checkpoint to make anything. You need a whole bunch of plumbing around it to make it usable by a developer. And so by the - I think - it was so obvious to me that a distribution platform like OpenRouter was critical to have in the ecosystem if we wanted there to be competition to Google. Like, unless-- ‘cause with Google, DeepMind is done training a new checkpoint, and then they push a button, and it gets blasted out across all their surfaces from Google Docs to,

Swyx [00:20:01]: Everywhere, even if I don’t want it.

Alex Atallah [00:20:02]: Everywhere. You wanna know about, like, on Android, like, overnight, they can deploy a new checkpoint to, like, a billion devices, right? And that invisible infra advantage, distribution advantage, most people don’t realize, but until OpenRouter showed up, - you had to think about all of that yourself as a model lab. And it was very daunting. at Anthropic, I think it took, well, more than twelve months to get to our first 10 million in revenue. And in contrast with Black Forest Labs, I remember the early days, you guys had a conversation with the BFL team, and, it was so simple for OpenRouter to say, “Oh, no problem. Like, the day you launch, we can send 1 million developers to you.” that was crazy. That was like a step function change in, like, an hour.

Swyx [00:20:46]: Is that a real number, a million?

Alex Atallah [00:20:47]: I,

Swyx [00:20:48]: Okay. All right.

Alex Atallah [00:20:48]: I think today it’s, like, 4 million. How many developers are on OpenRouter today?

Anjney Midha [00:20:52]: Over ten,

Alex Atallah [00:20:54]: Yeah.

Anjney Midha [00:20:54]: Over 10 million, but, like, it’s, it’s hard to, you

Alex Atallah [00:20:59]: I, yeah, I don’t know how to. Yeah.

Anjney Midha [00:21:00]: We do a lot of, like, account duping work, but, no

Alex Atallah [00:21:04]: If you could get 1,000 developers, just to put in context If you get 1,000 developers who try the model on day one after you release it and just, like, do inference and give you feedback, that’s a thousand

Anjney Midha [00:21:15]: That’s huge

Alex Atallah [00:21:16]: More developers than they knew how to get to on their own.

Swyx [00:21:19]: Well, BFL had a reputation, but yes.

Alex Atallah [00:21:21]: They had one in Stable Diffusion.

Swyx [00:21:22]: Yeah.

Alex Atallah [00:21:23]: And with Mistral, I don’t know if you guys remember, but the first checkpoint they released was, like, torrents. It was, like, torrent weights.

Swyx [00:21:31]: Yeah, they just put up a magnet link.

Alex Atallah [00:21:33]: Yeah, there was no API.

Anjney Midha [00:21:34]: Yeah.

Alex Atallah [00:21:34]: Because they didn’- they weren’t infra people.

Alex Atallah [00:21:37]: ? Like, it’s like, okay, download these weights, and you guys go figure out how to host it.

Swyx [00:21:39]: Well, he has a story on his side, yeah.

Anjney Midha [00:21:41]: Yeah, in addition to the, like, building a really good developer experience around it, the marketing that we do on, like, for different models is totally different and perceived totally differently

Alex Atallah [00:21:54]: Right

Anjney Midha [00:21:54]: From the marketing that a model lab does for itself.

Alex Atallah [00:21:56]: Yes, 1,000%.

Anjney Midha [00:21:57]: Right? We are like a, neutral layer looking at this market like it’s a big dark room with all the corners completely obscure to users, and users are walking into the room and, like, feeling around

Alex Atallah [00:22:09]: Yeah

Anjney Midha [00:22:09]: And trying to figure out what objects to grab off the tables and, like, build into, their companies. And it’s just an insane way of working. Like, models are not products where you can just enumerate all their features onto a web page. They’re all black boxes, including the open weight ones. So you need to, like, shine lights on all corners of this room, so that people can see what makes this model good, and you need the company shining that light to be a neutral third party, which is what we specialize in. So the, like. It’- In addition to developer experience, there’s also, like, a very important, like, marketing and product packaging component

Alex Atallah [00:22:50]: Yeah

Anjney Midha [00:22:50]: And a way of, like, routing and discovering models becomes, like, critical to your market as a provider or a model lab or a server tool and more in the future.

“Just a Wrapper”: Why VCs Misunderstood OpenRouter

Alex Atallah [00:23:03]: And this value, to your earlier point about how many VCs, like, just don’t. One of my biggest frustrations is that venture capitalists, many of them, like, just don’t have any operating experience in the field. so unlike a traditional investor who’s just maybe come up through the ranks as, like, a associate working on financial modeling or maybe hasn’t been a real operator in the field for, like, more than ten years, which is a big part of the industry now, I had just arrived at a16z, like, a year after running the platform. And so I knew what the challenges were of, like, building a real - great developer experience and like, being able to create a working piece of software with a model. And there were a few, I won’t name names, but there were investors who were looking at OpenRouter, and, felt at the time, like, when I would compare notes with people, that it was just, I quote unquote, “just a marketplace.”

Swyx [00:23:59]: Yeah, just a thin layer, just a

Alex Atallah [00:24:00]: Correct

Swyx [00:24:00]: Just

Alex Atallah [00:24:01]: A wrapper or whatever on other people’s APIs. And I was like, “You have no idea how strategic the value that OpenRouter has created by being able to orchestrate even three.” APIs in production. The amount of both engineering work and community design that goes into getting that live and running in production at the scale the OpenRouter team had started just doesn’t happen by default. And that was one of the things that stood out to me about Alex from the earliest days. Like, he just understood, like, - from a systems perspective, like, how do you get these flywheels going? Like, that stood out to me with OpenSea when we were working together on the NFT integration at Discord. Like, Alex had a level of community-- like, systems thinking on how you get these flywheels going that most scientists and machine learning people just don’t

Alex Atallah [00:24:48]: Think of. Like, we often think in terms of training.

Swyx [00:24:52]: It’s a linear stage.

Alex Atallah [00:24:53]: It’s this linear pipeline.

Swyx [00:24:53]: There’s no loop yet.

Alex Atallah [00:24:54]: Yeah. It wasn’t until much later that the modern context feedback loop cycle really got standardized in the industry. But at the time, if you remember, machine learning was like. Like, mostly we did a lot of ML, like, when I was in grad school on a laptop. So you just, like, download a dataset, ran some ablations, and you looked at the loss curves, and you’re like, “Great, I made AI.” And the idea that you have to, like, deploy those capabilities, collect feedback trajectories, then, like, put those into a continuous loop, like, came much later. And it was very counterintuitive to the - like, the traditional AI mindset. I do remember doing the investment phase for, OpenRouter, I just didn’t try and educate a bunch of other VCs on why it was not just a marketplace. I was like, “ what? I’m just gonna invest.”

Anjney Midha [00:25:41]: Yeah.

Alex Atallah [00:25:41]: And I’m going to, like, take the opportunity to partner with Alex, and if - no other VCs get it, that’s totally fine. ‘Cause at the time, - it was not obvious, I think, to several of the investors that, like, OpenRouter was not more than just a wrapper around APIs. And - that infuriated me. And I was like, “ what? I don’t have time to debate you. I’m - we’re gonna, we’re gonna invest.” And then I think, like, a month later, Matt Murphy marked it up by 10x. Like, - I think. I forget what the exact money was and so on, but, to his credit, Menlo Ventures realized, “Okay, there’s much more strategic value here as well.” Maybe you didn’t hear all these conversations behind the scenes But that frustrated me a lot. there’s a lot of this, like, opining about wrappers. and if you’re like, “Oh, an app is just a wrapper on a model,” then, like. And, OpenRouter is, like, this wrapper on top of other APIs, and this is the most stupid, reductive framework.

Alex Atallah [00:26:31]: And so it’s clearly somebody who has no experience deploying product at scale.

Swyx [00:26:34]: It’s the thing you dismiss other things with. Like, you’re a - everyone’s a wrapper on everything, right? Like, and there’s, there’s some Some wrappers have value.

Alex Atallah [00:26:40]: Investors are wrappers and LPs, right?

Alex Atallah [00:26:42]: Like venture capitalists. So, yeah, it’s all wrappers down, all down to bare metal, I guess, and like energy.

Swyx [00:26:46]: Yeah, there - When I started the whole AI engineer, I guess, the coining, in 2023, like, that was, like, the number one pushback is that this is no value. You should just train models.

Anjney Midha [00:26:56]: Right.

Swyx [00:26:57]: And, yeah, obviously this is, like. you guys are one of the testaments to the fact that you can build very valuable wrappers, but also very valuable model companies.

Alex Atallah [00:27:06]: It’s so, hard to be. Like, the day a model launches, the fact that you have an OpenRouter, endpoint for that model frequently at the top of Hacker News on day one, people don’t realize the amount of work that goes into accomplishing that. And OpenRouter used. Like, that would happen over and over again, and I remember going, “People have no idea how hard that is.”

Alex Atallah [00:27:30]: That’s not.

Swyx [00:27:31]: Yeah, we’ve covered some of the inference engineering that goes behind,

Alex Atallah [00:27:34]: Yes

Swyx [00:27:34]: Some of - with Base Ten and all those. Well, today you have, all those, like, cool code name things that people guess what Oxy Alpha is and all those things. But, like, I guess one of the things that you’re teasing is, how do you get that initial flywheel going, right? Because today you have your scale and your reputation, all these things, so obviously you - you’re driving immense distribution. But when you were early on, when it’s mostly

Bootstrapping OpenRouter Through Community

Alex Atallah [00:27:55]: The bootstrap, yeah.

Swyx [00:27:56]: Yeah.

Alex Atallah [00:27:56]: What was the bootstrap like?

Anjney Midha [00:27:58]: To bring it back to early Discord days, I think we, like, initially connected with. This is an OpenSea story, technically. But, and we initially connected when you were at Discord, and we talked about, like, - the Axie Infinity server.

Alex Atallah [00:28:13]: Oh, yes. Yes.

Anjney Midha [00:28:14]: This server was, like, the biggest server at the

Alex Atallah [00:28:17]: Yeah

Anjney Midha [00:28:17]: At Discord.

Alex Atallah [00:28:18]: That’s right.

Anjney Midha [00:28:19]: And you were like, constantly bumping up the

Alex Atallah [00:28:22]: The limits on the server. Oh, my God

Anjney Midha [00:28:24]: Of how many people could be in the server.

Swyx [00:28:24]: For those who don’t know, like, 10% of Philippines was Axie.

Alex Atallah [00:28:29]: Was on that server. That’s a big hit.

Swyx [00:28:31]: It was, like, a meaningful contributor to the GDP of the country.

Alex Atallah [00:28:33]: It was an NFT, like, crypto game, but it

Swyx [00:28:35]: It was like a Pokémon breeding thing.

Anjney Midha [00:28:36]: Yeah.

Alex Atallah [00:28:36]: Yeah. Similar. Yeah. There was battling, there was breeding, and then there was, like, a marketplace for trading.

Swyx [00:28:43]: Earn as well.

Alex Atallah [00:28:45]: Yeah, earn. And, like, the graphics were really cute and fun, and you like, you get emotional about your Axie that you make. So to, like, start a community like that, which we had to do many times at OpenSea with every early project, for us to create a marketplace for it, we need to make sure that the, like, the community wants it.

Anjney Midha [00:29:09]: Right.

Alex Atallah [00:29:09]: And it’s like building something that people want and going and telling them about it. Like, you can do that on a one basis, but there’s way higher leverage to do that in a community where everyone can talk to you at the same time. So we spent a lot of time, like, building things that the community really wanted. We did the same thing for OpenRouter. And, like, the Axie community was one of, like, a zillion communities we did that with. And Anj, like, saw us doing it and. ‘Cause you could just see people sharing OpenSea links constantly in that Discord. Like, users sharing links is a really clear indicator that, like, something important is going on. So we spent, a lot of time, like, first figuring out what the gap is in the technology that people care about. Like, what was the actual problem that needs to be solved? in early LLM days, it was, OpenAI refusing to finish the prompt or,

Anjney Midha [00:30:09]: Yeah

Alex Atallah [00:30:10]: To, like, complete the task. It was also.

Anjney Midha [00:30:13]: Inability to customize models. and so there are communities that, like are just completely blocked on that issue, and those are the communities that are most useful to learn about and dive into and explore.

Alex Atallah [00:30:28]: Something that really struck me at that time, - as I was just hearing your talk, I remember noting - you may not remember this, but we - we had these, like working, Zoom calls that we were doing a sprint around for, like this OpenSea integration with Discord. and, we’d, we’d - it was myself, my engineering team. I think you were there. And I remember, Alex, in the middle of one of those calls, just like there was like silence. we were all like, “Oh, yeah, this totally makes sense. Let’s do this.” And then there’s - every, like everybody aligned. And Alex was like, “No, this makes no sense to me.” And everyone’s - I remember going, “What? Like, it works. Like, you click on a link and this, then it bounces you out to, like, OpenSea.” And he was like, “It’s not a good user experience. Yeah, we should not do this.” And I remember going, he was the only one person out of all of us to raise his hand and go, yes, it made sense from a technical implementation perspective. Like, we were bouncing the user out into the, into OpenSea. And so it kinda checked the box of the product manager’s requirements on both sides. But Alex went one step further and was like, “ what would be better, guys? If we just embedded the experience right here inside of Discord so the link opened up as an embedded iframe, and you can just check out right there.”

Alex Atallah [00:31:47]: And not one person on the call, and there’s like seven of us who had met, like, week after week.

Swyx [00:31:52]: And it’s the guy who doesn’t work for Discord.

Alex Atallah [00:31:53]: And it’s the guy who doesn’t work for Discord.

Swyx [00:31:55]: Like, technically, you benefit if they bounce.

Alex Atallah [00:31:57]: Exactly. And that was, like, adversarial. To keep the user inside of Discord would be adversarial to OpenSea. And yet Alex put that user experience first. And I was like, “That’s special.”

Swyx [00:32:08]: Wow.

Alex Atallah [00:32:08]: Because it’s very hard to have somebody who’s technical like Alex and understands the developer flow, but also understands the best user experience and wants to prioritize that. And that’s two sides of the flywheel that if you can get spinning, like is often hard to stop. And you just reminded me, like that one was one of those moments where I go, I - I realized I gotta be better at user experience because I should have been the one who came up with that, and I didn’t. And I learned from you. And, I think that went into one of our case studies for the PM training program at Discord.

Swyx [00:32:34]: Whoa.

Alex Atallah [00:32:36]: I don’t know if it there is Because of

Swyx [00:32:38]: You need an Alex is the conclusion.

Alex Atallah [00:32:40]: Yeah. You need an Alex. And this is why I’m not, nobody should be surprised why Stripe decided like they had to buy OpenRouter because it’s a really rare combination of people who understand the machine learning community, the developer experience, and the user experience. And putting all that together has resulted in this extraordinary scale that very few other marketplaces have been able to achieve

Window AI, BYOM, and Finding the Right Form Factor

Swyx [00:33:02]: Yeah.

Alex Atallah [00:33:02]: Over the last, five years.

Swyx [00:33:04]: Yeah. Well, we should talk about the other reasons for acquisitions, which

Alex Atallah [00:33:07]: Yes, we should.

Swyx [00:33:07]: You’ve written about. I wanna proceed somewhat chronologically as well. So - there is a point that, one of the questions that, Dave from H of Zero sent in was, when did it - really started to work? And you brought up Mixtral. I don’t know if you wanna bring up that story.

Alex Atallah [00:33:22]: Oh, yeah.

Swyx [00:33:23]: Which obviously you overlap with, so.

Anjney Midha [00:33:26]: Yeah, the MoE was. I don’t know when. there’s no like one moment where I was like, “Oh, this is, officially starting to work.” It was

Swyx [00:33:36]: The moment where you had a Chrome extension, like, really super early on.

Anjney Midha [00:33:39]: Oh, yeah. But, well, - yeah. So before OpenRouter, I wanted to, like, explore a bring-your-own-model experiment. And,

Swyx [00:33:47]: Which anyone familiar with crypto is like, yeah, Phantom and all these things.

Anjney Midha [00:33:50]: Yeah. So it felt like doing a MetaMask analogy for AI would be a fun way of exploring that. And at the time, there were no AI apps. There were probably as many AI apps that were, like, hitting AI - like, hitting an LLM via an API call as there were, like, games just doing it in JavaScript. like there was a, there was a moment in time where it could have been the case that web apps call LLMs through the browser, like through some desktop

Alex Atallah [00:34:27]: Yes.

Anjney Midha [00:34:27]: Managed app that is controlled by the user. and of course, there are like, I think, many reasons that did not happen. But back when the days were that primordial, I built a Chrome extension called Window AI

Swyx [00:34:43]: With Plasmo.

Anjney Midha [00:34:44]: With Plasmo.

Swyx [00:34:45]: I had come across early on, and I was like, “Who’s gonna use this?” You did.

Anjney Midha [00:34:49]: Plasmo had a couple, like, I think Phantom was using it. there were some other, like real companies using it.

Alex Atallah [00:34:56]: It was like a shim.

Swyx [00:34:57]: React for Chrome extension. It compiles to all

Anjney Midha [00:35:00]: Yeah.

Alex Atallah [00:35:00]: I see.

Anjney Midha [00:35:00]: Like Next.js for Chrome extensions.

Swyx [00:35:01]: Next.js, Next.js.

Alex Atallah [00:35:02]: Okay.

Anjney Midha [00:35:03]: And yeah, built Window AI on top of it. The creator of Plasmo, like started contributing code to Window AI, in GitHub, and that turned out to be Louis Vicchi

Alex Atallah [00:35:15]: Oh, you’

Anjney Midha [00:35:15]: Who is the founder of OpenRouter.

Alex Atallah [00:35:17]: That’s right. You have told me this is how you met Louis. Yes.

Anjney Midha [00:35:19]: Yeah.

Alex Atallah [00:35:19]: Okay.

Anjney Midha [00:35:20]: So, that allowed users to like configure which model they wanted to use for a web page in their browser, and then, like the app would just call out to that model when it needed to do things. not the right form factor for LLMs, but, it’s like fun experiment. You learn a lot, and like I open sourced it. And the main learning is like, okay, this has to be an API, and it has to look a little bit - like, there has to be more of a developer experience here and more of a discovery experience as well. Like, I don’t know where to use these models, and a little Chrome extension is not gonna help me discover. It’s not enough real estate. I need more space. I need visuals. I need graphs. I need, examples. I need images. I need to, like, I need to be able to, like explore both as a human and as an agent.

Crypto, Midjourney, and the Early Generative AI Ecosystem

Alex Atallah [00:36:10]: Yeah.

Anjney Midha [00:36:10]: So that’s how OpenRouter came to be.

Alex Atallah [00:36:13]: A meta point that.

Alex Atallah [00:36:16]: I think is underappreciated, but Alex is reminding me, is that we were quite lucky that we were so. we were, like, adjacent to the crypto community in those days. Because in hindsight, crypto ended up being like a dress rehearsal for generative models, right? If you think about the Axie experience, Alex is totally right, there were not that many AI apps at the time. And while I was dealing-- my job was to be the head of platform at Discord, which meant to be a general purpose place for communities and friends to create-- for developers to create apps and bots and, other services that could be deployed across Discord. And while 80% of the attention at the time was being spent on crypto, because that’s where all the NFT volume was, there was, like, twenty percent of my time I was spending with a friend, who would get hotbot with me and ask me for. We would play Magic: The Gathering on weekends, and he was working on a little Discord bot that could take a text input and turn it into an image, and it was called Midjourney. You

Swyx [00:37:15]: Is that David?

Alex Atallah [00:37:15]: It was David Holz.

Alex Atallah [00:37:16]: He was a good friend. And David and I have both been failed ARVR founders, in the before that. And, I remember this. Midjourney was one of the fastest-growing communities we had after Axie Infinity started to peter off. And many of the, like, the abstractions and the infrastructure decisions we made to scale Axie happened just in time because they. Axie did this and then fell off a cliff. And then as Midjourney was taking off, we, like, explicitly decided to help David make the server, the Midjourney server, as the primary place for interaction with the model, because it was very hard for people to understand how to use the model if they couldn’t see other people using it and copy them. And so the single-player Midjourney web app on its own, like midjourney.com, had, like, terrible retention because people would show up, they’d see this empty field. It’s like E 2, and they would type in, like, cat or dog. And it was, like, paralyzing for them to have this blank canvas that they had to fill because they’d never used an AI model before. But instead, in a Discord server, you could see other people using it and riff off of their prompt, and the engagement was off the charts. And so scaling, Midjourney from zero to, like, 10 million monthly actives was a much smoother approach Axie Infinity. And so,

Swyx [00:38:29]: Don’t forget the best of four pictures, and you choose one.

Alex Atallah [00:38:31]: The best, yeah, and then the other, we

Swyx [00:38:32]: Which is the feedback loop.

Alex Atallah [00:38:33]: The RLHF feedback loop, which, by the way, separately, like, Tom Brown, David and I used to play Magic: The Gathering on weekends. And so, like, it was one group of friends would hang out, and we’d. Like, these concepts were all being discussed all the time. But, there was.

Alex Atallah [00:38:47]: I think there were few of us who bridged both the crypto worlds and the AI worlds. And compared to crypto, where it was - the question was always, what’s the use case, for this technology? There was never any need to ask that for AI because it’s, like, the use case was so visceral. It was like, I can create now anything at - I can imagine. I can write novels, I can code. And the infrastructure that those of us who believed in the distributed systems, like, value of crypto, like the censorship resistance part, found this use case that was explosive. And I think between Midjourney, the, Claude was a Discord bot launch, that we were using internally as an LLM. ElevenLabs had a TTS model that we had on Discord as well. Like, Discord became this petri dish for, like, early apps to innovate. And I don’t think it’s a coincidence that they found a home there before OpenRouter gave the world, like, a public home store or, like, a, storefront. Discord was this, like, almost petri dish storefront that - had, like, piggybacked on the infra we’d built for crypto communities. And then I think Alex was one of the first people to realize, wait a minute, like, these apps need their own home, on the internet. And then OpenRouter, to me, was a continuation of that community’s needs. And of course, there was the crazy distribution that you enabled for a lot of these developers.

Why OpenRouter Couldn’t Just Live Inside Discord

Swyx [00:40:07]: So then my question is, how come you were. My perception is OpenRouter is not that Discord-centric, right? You have a Discord.

Anjney Midha [00:40:14]: Yeah.

Swyx [00:40:14]: And you use it to engage your community, but it’s not like Midjourney where, like, no, that is like the primary way people experience OpenRouter.

Anjney Midha [00:40:21]: Yeah, Midjourney, like, it really helps to see visually really quickly how people are using the model and how to prompt it.

Swyx [00:40:29]: Yeah.

Anjney Midha [00:40:29]: And I think that is partly why the server was so critical. It’s like it is the user experience. It adds a ton.

Swyx [00:40:36]: Yes.

Anjney Midha [00:40:37]: And you can go the whole mile with just, like, prompting via Midjourney, like, the, via the Midjourney Discord server, getting your images and then sharing them and having fun. For OpenRouter, for LLMs, like, you need a lot of user experience around LLMs to make them, like, really usable.

Swyx [00:40:54]: Charge point.

Anjney Midha [00:40:55]: And yeah.

Anjney Midha [00:40:57]: The, like, seeing the examples of other people is also not as useful because it’s a lot of stuff to read. It takes a long time.

Swyx [00:41:03]: Yeah.

Anjney Midha [00:41:04]: You need, like, based integration. Not possible to do in a Discord server. You need, Or technic- it’s possible. I shouldn’t say that. It’s just not a great developer experience. you need, like, - you need governance for. At the point where you got based integration, now you need governance for managing the LLMs that have access to it, the data policies, which teams. All that stuff needs a lot more than a Discord server can provide. So it’s just

Swyx [00:41:30]: Yeah

Anjney Midha [00:41:30]: It’s not the right.

Alex Atallah [00:41:32]: Well, in addition, you’re not wrong, but also there’s the very important distinction that, Midjourney was an end user application.

Swyx [00:41:40]: Right.

Alex Atallah [00:41:40]: And, that’s why Discord, which has 250 million monthly end consumers, made, it made sense for Discord to be a host for that application experience. What I knew was gonna happen soon after Midjourney found explosive product-market fit, because we. I think when Midjourney launched, from launch to $100 million revenue run rate, it was less than eight months. And shortly thereafter, Stable Diffusion launched. And, all of us used to hang out in the Discord server. There, I think it was the,

Swyx [00:42:13]: The Stability Discord?

Alex Atallah [00:42:14]: It was the

Swyx [00:42:16]: Yeah, LAION.

Alex Atallah [00:42:16]: Yeah, the LAION Discord server.

Swyx [00:42:17]: The image community that spawned Stable Diffusion.

Alex Atallah [00:42:19]: The image community. Yeah. And so when Stable Diffusion came out, I realized- Oh, now other people can build their own Midjourney.

Alex Atallah [00:42:27]: Because until then, Midjourney did not have an API, so they were a stack company, right? They were training their own models, and they were deploying them as an application. But if you wanted to build your own Midjourney, there was no API of that quality. and I think E two was still quite primitive. Like, Midjourney had great quality. And then when Stable Diffusion came out, suddenly there was this new person who - there was - this new capability in the world, which is a developer could create their own Midjourney. And that, I think, created the need for something like OpenRouter, because then you need an API to. If you - if you had the creativity of David Holz and you had Stable Diffusion as the model and you wanted to put these things together, how could you do that without having to figure out how to host the weights? And what OpenRouter, - the shape of OpenRouter enabled is that. Right? When you have open model alternatives to closed applications, OpenRouter’s value in the world becomes extraordinary because now any developer can just show up and use the

Stable Diffusion and the Need for a Model API Layer

Swyx [00:43:20]: You just love model diversity.

Anjney Midha [00:43:21]: Did you just say the shape of OpenRouter?

Alex Atallah [00:43:23]: Oh, no.

Anjney Midha [00:43:25]: Were you in cloud? What is this the real Han?

Alex Atallah [00:43:26]: I’ve been, I’ve been - I’m, I’m misaligned now. I’ve been overtrained. I’ve been using Cloud way too much, haven’t I?

Swyx [00:43:34]: Claude-ish is what people would say.

Alex Atallah [00:43:35]: Claude-ish. Oh, God, I gotta untrain myself.

Swyx [00:43:38]: Okay. - And I just wanna cap off the Mistral side. my TLDR is there was a Mistral price war, is what they called it, right? Like, round about NeurIPS is twenty-three or twenty-four.

Mistral and the Birth of the Inference Marketplace

Anjney Midha [00:43:47]: Yes. December

Swyx [00:43:48]: They launched, the Mistral 8x7B, and like the price went down like 80%.

Anjney Midha [00:43:54]: Yeah.

Swyx [00:43:54]: To me, that’s very positive because it’s like the first, like, real competition to host Mistral. Is there more?

Anjney Midha [00:44:01]: Yeah, that was. I’m, like, trying to remember it, all the things that happened. It. Like, we saw that model come out and immediately saw people say that it was the best model in the world.

Alex Atallah [00:44:15]: Yes.

Anjney Midha [00:44:15]: Like, this was, to my knowledge, the first time an open weights model was called that in real seriousness.

Swyx [00:44:22]: It’s hype, right? Is it?

Anjney Midha [00:44:25]: It was hype. It was hype. It was also, like, hype from AI influencers at the time. And there were many examples where it was, like, outperforming four. So people really wanted to try it out and see, is this gonna be true for me too? And if so, at what price? And, the, like, inference landscape was really messy.

Alex Atallah [00:44:49]: Yes.

Anjney Midha [00:44:50]: We cleaned it up. - it allowed, like, providers to compete on price, so we could give you just the best price in one spot. And so it was, I think, the first clear example of, like, a provider marketplace working in a way that adds value to end developers.

Alex Atallah [00:45:08]: Sean, you may not remember this, but I think we met for the first time a few days after Mistral came out at NeurIPS

Anjney Midha [00:45:15]: Yeah.

Alex Atallah [00:45:15]: At a luncheon.

Swyx [00:45:16]: Yeah. That’s where I also met BFL as well. Yeah.

Alex Atallah [00:45:18]: And Guillaume was there.

Swyx [00:45:19]: Yeah.

Anjney Midha [00:45:19]: I was at NeurIPS at that time.

Alex Atallah [00:45:20]: You were there too. And, we had just announced the Mistral investment, and I remember Guillaume was over there, and I remember turning to Guillaume and asking him, Like, “Is it is all the. Like, how are you feeling after the launch of Mistral and seven B?” And, him in his typical French fashion was like, “ it’s a, it’s an okay model. It’s not that good.” And I was like. It was so, in contrast. But I remember him also saying that part of the reason he felt a lot of people Thought that it was better than four was because of the speed. - it was an MoE model that they had, like, absolutely figured out how to make super efficient. It was on the Pareto frontier. And this is an important thing about LLMs, right? Sometimes when they’re faster, you think they’re smarter, even though, like, if you did, N of, these common, like, evals that are - you do seven tries, and I don’t remember. I think we should go back and figure out what the data says, but I wouldn’t be surprised if it turns out, oh, on an N of seven attempts, four was smarter on evals, but the perception of on, like, or correctness would be smarter or more accurate. But, people, like, from a human preference perspective felt that it was faster because it - or smarter because it’s so fast.

Swyx [00:46:36]: Yeah. And most queries do not take that level

Alex Atallah [00:46:39]: Don’t take that. That’s true.

Swyx [00:46:40]: Right? So this is the start of humans as router

Alex Atallah [00:46:42]: Yes.

Swyx [00:46:42]: Which then eventually becomes OpenRouter as router of like the

Alex Atallah [00:46:45]: Oh, that’s interesting way to think about it. Yeah.

Swyx [00:46:47]: Like, because humans are the routing mechanism. Like, I will ask the fast model first, and then if, like, oh, not good enough, I’m gonna upgrade manually.

Alex Atallah [00:46:52]: Yes.

Swyx [00:46:53]: But then he’s gonna auto it.

Alex Atallah [00:46:54]: I didn’t, I hadn’t thought of it that way, but that makes sense.

Swyx [00:46:57]: Which then there’s, there’s a lot more techniques, like fusion. Fusion is the thing that we should talk about. Before I move on to those things, I just want to close off the early years. one thing that I observe, which you are also an investor in Arena.

OpenRouter vs. LM Arena

Alex Atallah [00:47:10]: Right.

Swyx [00:47:10]: And we talked about Midjourney having that feedback loop of, A, B, C, D, and choosing that very. being very important. And you understand the flywheel. So how come you didn’t build Arena, and how come Arena didn’t build OpenRouter?

Anjney Midha [00:47:23]: Well, Arena started before OpenRouter, right?

Swyx [00:47:27]: They had the school project

Anjney Midha [00:47:29]: Yeah, LM

Swyx [00:47:29]: And then it became a company.

Anjney Midha [00:47:31]: LM Arena, yeah.

Swyx [00:47:32]: So, but, and I know you had some Arena experiences, like the up comparison type things.

Anjney Midha [00:47:37]: Yeah.

Swyx [00:47:37]: But you never really went as hard as Arena did.

Swyx [00:47:40]: And,

Anjney Midha [00:47:40]: In doing up experiences?

Swyx [00:47:42]: Yes. And LM Arena did have a router project based on LM Arena ELOs, which they never commercialized.

Anjney Midha [00:47:48]: It’s hard to do a company that does both because one company is taking data and selling it, and the other company really can’t by default. So, I think there is, like, a branding reason that there are two companies here. like, when you set up OpenRouter, there’s no training, there are no prompts, right, aside from what your provider policy set. Like, OpenRou- like, OpenRouter can’t see your prompts or completions. If you want to see that as an org, you have to opt into it and enable it. And so we’re, like, pretty conservative and careful about data policy and security. And privacy. And LM Arena is like, their business model is like oriented around the labs and,

Swyx [00:48:34]: Because they give it for free, right? You don’t give it for free to give it for free.

Anjney Midha [00:48:37]: Yeah.

Anjney Midha [00:48:38]: But we do give some. We like have free endpoints too, but like those free endpoints, we, I think we’re not collecting any prompts. We’re not like monetizing the data unless you, opt into it for some reason.

Alex Atallah [00:48:48]: This comparison. you’re not the first person to ask me this, and Alex knows this, but I was the interim, like the founder, like first CEO of Arena for the first five months when, and we were helping Anastasios and Waylin spin out of Berkeley. And, I did invest in that before, OpenRouter, but it was very strange to me the comparisons that outside, folks would make between the two projects because the missions were completely different. The founding entity for Arena, we called it the AI Reliability Institute because it was there as an eval service. Like the data, so to speak, that they were originally, offering the labs was how do you make the evaluation of models more reliable than like the state of the art at the time, which was like really just finger in the wind.

Alex Atallah [00:49:38]: That’s what Anastasios and Waylin’s PhD work was as scientists at Berkeley, was on statistical methodologies for correcting, eval estimates, based on like intrinsic biases and how you collected the data.

Swyx [00:49:54]: Yes.

Alex Atallah [00:49:54]: And

Swyx [00:49:54]: Style control.

Alex Atallah [00:49:55]: Style control and stuff like that. And which is very much like a, hey, how. If you’re a scientist and you’re trying to. the highest expectation customer for Arena was always like a training and, like a researcher at a lab. Whereas the highest expectation customer from my perspective that Alex like really understood and was the mission was to serve was like a developer, right? Who then takes the result of the research and then produces an application that’s deployed to the world. It was a completely different problem and person that these two teams were focused on. And so from the outside in. I don’t know if you remember this, but I have a distinct memory of a few weeks before we did the term sheet, together for OpenRouter, I’d given you a call because we were trying to get a pooled data set together from OpenRouter and from Arena to, create like an open source repository of prompts. these projects were so different in their goals that it was totally normal to me to be like, “Oh, yeah, let’s call Alex and see if he’d want to team up on pooling data,” because they’re so different. We need. We don’t have that data at all. We. Like, we didn’t have API prompts. We didn’t, we didn’t have like what developers want to do with the models, which is very different from what researchers inside a model lab want to do before releasing the model.

Swyx […