Podcast
Deep Values Frontier: AI Research Explained
AI explained from first principles: new AI research papers, large language models (LLMs), reasoning, memory, AI agents, AI chips and the companies building them. Each episode turns one real source, a paper, talk or interview from a leading lab, into a clear two-voice conversation with the source linked. From Deep Values (deepvalues.ai), the team behind the Deep Value Intelligence multi-agent analyst and the open-source research at research.deepvalues.ai. Chinese edition: 深度價值前沿. Video on YouTube @deep-values. Also on Apple Podcasts. The hosts are AI-generated voices, not the people discussed.
- Episodes
- 8
- Releases
- Daily
- Typical length
- 21 min
- On air since
- 2026
- Latest episode
- 19 hours ago
Episodes
8 episodes, newest first
-
18 min
Cursor Engineer Ships 1,000 PRs a Month. Who Checks Them? | AI Explained
Show notes
Learn the guardrails that took Cursor engineer Lauren Tan from babysitting a few AI agents to 1,000 pull requests in a month. It matters now because unchecked vibe coding is turning fast code into endless debugging.
Chapters
(0:00) Scaling AI Software Agents
(4:01) Step 4: Autonomous Cloud Orchestration
(8:28) Step 2: Unit Testing Agents
(10:28) Evolution of Engineering Scale
(11:08) Step 3: Extreme Systemic Guarantees
(14:33) Cost of Autonomy: Trade-Offs
(15:15) Democratizing Code for Teams
(15:52) Frontier of Autonomous Agents
(17:35) Apps for Intelligent Investors
(17:53) Get the free apps
DeepValues toolkit
• Run the numbers on any stock yourself, free: https://app.deepvalues.ai/?ref=podcast&src=how_engineers_build_autonomous_ai_coding_age-en
• Free Margin of Safety Weekly, every Saturday (guru moves, market temperature, the week's research; no account needed): https://app.deepvalues.ai/letter?src=spotify&ref=how_engineers_build_autonomous_ai_coding
• Open-source research on 300+ companies: https://research.deepvalues.ai/
• DeepValues for iOS, intrinsic value and margin of safety for any stock: https://apps.apple.com/us/app/deepvalues/id6775568608
• DeepValues for Android: https://play.google.com/store/apps/details?id=app.deepvalues.android&referrer=utm_source%3Dspotify%26utm_medium%3Ddescription%26utm_campaign%3Dai
• Is the market overvalued? Track it with Buffett Indicator (iOS): https://apps.apple.com/us/app/buffett-indicator/id6747404614
• The same market gauge on Android, Bubble Watch: https://play.google.com/store/apps/details?id=com.luluventures.buffett&referrer=utm_source%3Dspotify%26utm_medium%3Ddescription%26utm_campaign%3Dai
• Deep Value Intelligence, the multi-agent AI analyst: https://deepvalues.ai/
• Developer API for your AI agents: https://app.deepvalues.ai/developers
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors. Understand AI from first principles.
• This episode on YouTube: https://youtu.be/tVFul_AwDV4
-
19 min
Supabase Says 70% of Its New Users Aren't Human | AI Explained
Show notes
Learn why 70% of Supabase's new users are AI agents, not people. Software was long gated by human engineering hours, and that barrier is falling now.
Chapters
(0:00) The AI Coding Era
(3:52) Infrastructure is Already Here
(4:33) Operating at Machine Speed
(5:13) Tipping Point of Autonomous Activity
(6:14) Platforms Must Rebuild for Non-Humans
(6:56) Open Questions of the Agentic Era
(12:41) Rise of Personal Software
(15:41) Vision Overtakes Execution
(17:46) Rise of Personal Software
(18:30) Apps for Intelligent Value Investors
(18:42) Get the free apps
DeepValues toolkit
• Free Margin of Safety Weekly, every Saturday (guru moves, market temperature, the week's research; no account needed): https://app.deepvalues.ai/letter?src=spotify&ref=why_most_supabase_users_are_ai
• Open-source research on 300+ companies: https://research.deepvalues.ai/
• DeepValues for iOS, intrinsic value and margin of safety for any stock: https://apps.apple.com/us/app/deepvalues/id6775568608
• Is the market overvalued? Track it with Buffett Indicator (iOS): https://apps.apple.com/us/app/buffett-indicator/id6747404614
• The same market gauge on Android, Bubble Watch: https://play.google.com/store/apps/details?id=com.luluventures.buffett&referrer=utm_source%3Dspotify%26utm_medium%3Ddescription%26utm_campaign%3Dai
• Deep Value Intelligence, the multi-agent AI analyst: https://deepvalues.ai/
• Developer API for your AI agents: https://app.deepvalues.ai/developers
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors. Understand AI from first principles.
• This episode on YouTube: https://youtu.be/rCLaElikewU
Subscribe: https://youtube.com/@deep-values?sub_confirmation=1
Podcast (Deep Values Frontier): Spotify https://open.spotify.com/show/2FlcJXyfpOMhbSFhxYRSqu · Apple Podcasts https://podcasts.apple.com/us/podcast/id6826607834
Investing channel: https://youtube.com/@TheDeepValues
The two hosts' voices are AI-generated from the sources discussed. Educational content only; not investment advice.
#AI #ArtificialIntelligence #LLM
-
22 min
SpaceX Engineer: Always-On AI Agents Hit a 200K Token Wall
Show notes
A SpaceX engineer's Stanford lecture shows why AI agents stall at a 200,000-token memory wall, and why the model isn't the bottleneck. The harness around it now decides whether agents can work unattended for weeks.
Chapters
(0:00) AN ALWAYS-ON
(2:18) The architectural shift toward ambient
(4:44) The isolated cloud computer
(5:25) System Blueprint Step 3: Surviving the
(8:09) Structural
(8:48) Designing the cognitive memory mode1
(10:05) 10to 60 Minutes
(10:44) Designing the cognitive memory mode1
(12:02) Structural
(13:21) The Escalation Ladder
(15:06) System Diagnostics
(16:17) Assume the models today are
(17:54) System Diagnostics
(21:18) Apps for Intelligent
(21:52) Get the free apps
DeepValues toolkit
• Free Margin of Safety Weekly, every Saturday (guru moves, market temperature, the week's research; no account needed): https://app.deepvalues.ai/letter?src=spotify&ref=engineering_always_on_ai_agents
• Open-source research: https://research.deepvalues.ai/
• DeepValues for iOS, intrinsic value and margin of safety for any stock: https://apps.apple.com/us/app/deepvalues/id6775568608
• Is the market overvalued? Track it with Buffett Indicator (iOS): https://apps.apple.com/us/app/buffett-indicator/id6747404614
• The same market gauge on Android, Bubble Watch: https://play.google.com/store/apps/details?id=com.luluventures.buffett&referrer=utm_source%3Dspotify%26utm_medium%3Ddescription%26utm_campaign%3Dai
• Deep Value Intelligence, the multi-agent AI analyst: https://deepvalues.ai/
• Developer API for your AI agents: https://app.deepvalues.ai/developers
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors. Understand AI from first principles.
• This episode on YouTube: https://youtu.be/yXYfMl5_wLc
Subscribe: https://youtube.com/@deep-values?sub_confirmation=1
Podcast (Deep Values Frontier): Spotify https://open.spotify.com/show/2FlcJXyfpOMhbSFhxYRSqu · Apple Podcasts https://podcasts.apple.com/us/podcast/id6826607834
Investing channel: https://youtube.com/@TheDeepValues
-
22 min
Meme Coin Millions Funded AI Doomsday. Who Audits It?
Show notes
A meme-coin donation to the Future of Life Institute ballooned into hundreds of millions of dollars and turned a niche think tank into a force in AI policy. We audit Effective Altruism, P(doom) and Vitalik Buterin's d/acc as capital allocation decisions and look at what this ideological moat means for frontier AI capex.
Chapters
(0:00) The Ideological Ledger of AI
(2:52) Effective Altruism and Earning to Give
(3:35) Four Phases of EA Allocation
(4:41) Effective Altruism and Earning to Give
(5:21) Four Phases of EA Allocation
(7:13) Case Study: Vitalik's Dogecoin Windfall
(8:53) The Institutional Moat
(10:09) Four Phases of EA Allocation
(10:30) Effective Altruism and Earning to Give
(11:10) Macro Risks and the Bear Case
(12:55) The Institutional Moat
(13:32) Verdict: Ideology as a Risk Factor
(14:55) The Ideological Matrix: EA, e/acc, d/acc
(18:19) Verdict: Ideology as a Risk Factor
(21:01) Investor Apps and Resources
(21:20) Get the free apps
DeepValues toolkit
• Free Margin of Safety Weekly, every Saturday (guru moves, market temperature, the week's research; no account needed): https://app.deepvalues.ai/letter?src=spotify&ref=how_meme_coins_funded_ai_doomsday
• Open-source research on 300+ companies: https://research.deepvalues.ai/
• DeepValues for iOS, intrinsic value and margin of safety for any stock: https://apps.apple.com/us/app/deepvalues/id6775568608
• Is the market overvalued? Track it with Buffett Indicator (iOS): https://apps.apple.com/us/app/buffett-indicator/id6747404614
• The same market gauge on Android, Bubble Watch: https://play.google.com/store/apps/details?id=com.luluventures.buffett&referrer=utm_source%3Dspotify%26utm_medium%3Ddescription%26utm_campaign%3Dai
• Deep Value Intelligence, the multi-agent AI analyst: https://deepvalues.ai/
• Developer API for your AI agents: https://app.deepvalues.ai/developers
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors. Understand AI from first principles.
• This episode on YouTube: https://youtu.be/HQ_raMftJCA
Subscribe: https://youtube.com/@deep-values?sub_confirmation=1
-
19 min
Ilya Sutskever: AI Will Do Everything You Can in 3-10 Years
Show notes
Ilya Sutskever told University of Toronto graduates that the brain is a biological computer, so a digital one will eventually learn anything a human can. Here is his full logic, from AI doing its own research to progress going vertical, and why he says hands-on use of today's AI is the only way to build the intuition this calls for.
Chapters
(0:00) The Inevitable Future of AI
(2:17) A Decade at University of Toronto
(3:00) Psychological Friction of the Future
(5:19) The Capability Shift in Computing
(5:39) Current Deficiencies and Open Questions
(6:20) The Brain as Biological Computer
(9:16) The Logic of Extreme Acceleration
(9:56) Cascading Impacts of Autonomous AI
(10:59) Psychological Friction of the Future
(12:20) Building Intuition Through Observation
(14:02) Current Deficiencies and Open Questions
(16:05) Humanity's Greatest Challenge and Reward
(18:10) Apps for Intelligent Value Investors
(18:34) Get the free apps
DeepValues toolkit
• Open-source research on 300+ companies: https://research.deepvalues.ai/
• DeepValues for iOS, intrinsic value and margin of safety for any stock: https://apps.apple.com/us/app/deepvalues/id6775568608
• Is the market overvalued? Track it with Buffett Indicator (iOS): https://apps.apple.com/us/app/buffett-indicator/id6747404614
• The same market gauge on Android, Bubble Watch: https://play.google.com/store/apps/details?id=com.luluventures.buffett&referrer=utm_source%3Dspotify%26utm_medium%3Ddescription%26utm_campaign%3Dai
• Deep Value Intelligence, the multi-agent AI analyst: https://deepvalues.ai/
• Developer API for your AI agents: https://app.deepvalues.ai/developers
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors. Understand AI from first principles.
• This episode on YouTube: https://youtu.be/cP5A-izQqaE
Subscribe: https://youtube.com/@deep-values?sub_confirmation=1
Podcast (Deep Values Frontier): Spotify https://open.spotify.com/show/2FlcJXyfpOMhbSFhxYRSqu · Apple Podcasts https://podcasts.apple.com/us/podcast/id6826607834
Investing channel: https://youtube.com/@TheDeepValues
-
22 min
Why Google Wants Data Centers in Space: A 20-Year Bet | AI Explained
Show notes
Sundar Pichai explains why Google keeps giving scarce chips to frontier research ahead of paying cloud customers, and why it is putting TPUs into sun-synchronous orbit. The limit on AI is no longer the software. It is electricity, land, about 2,000 separate components and the human electricians needed to plug it all in.
Chapters
(0:00) Google's Infinite Compute Strategy
(2:18) Argon Launch: Scale and Impact
(3:39) Standard vs Frontier Models
(4:22) The RSI Engine
(6:23) Path to AGI: Converging Architectures
(9:27) Cloud Revenue vs Frontier R&D
(12:08) Safety and Supply Chain Bottlenecks
(12:51) Data Centers in Space
(15:09) Safety and Supply Chain Bottlenecks
(16:52) Data Centers in Space
(17:35) Computing Adapts to Humans
(19:01) The Headless Agent Architecture
(19:44) Argon Launch: Scale and Impact
(20:43) Infinite Compute and Persistent Agency
(21:28) Get the free apps
DeepValues toolkit
• Free research reports on 300+ companies: https://research.deepvalues.ai/
• Deep Value Intelligence, the multi-agent AI analyst: https://deepvalues.ai/
• Developer API for your AI agents: https://app.deepvalues.ai/developers
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors. Understand AI from first principles.
• This episode on YouTube: https://youtu.be/x_OiRc7wxmQ
Subscribe: https://youtube.com/@deep-values?sub_confirmation=1
Podcast (Deep Values Frontier): Spotify https://open.spotify.com/show/2FlcJXyfpOMhbSFhxYRSqu · Apple Podcasts https://podcasts.apple.com/us/podcast/id6826607834
Investing channel: https://youtube.com/@TheDeepValues
The two hosts' voices are AI-generated from the sources discussed. Educational content only; not investment advice.
#AI #ArtificialIntelligence #LLM
-
21 min
Why AI Must Learn to Forget: DeepMind's 1M-Token Problem
Show notes
Google DeepMind research VP Ed Chi explains why the next AI assistant will be judged on its worst day, not its smartest, and why memory must rank and forget rather than store everything. The moat is shifting from clever models to capability per dollar, and that changes how you read every AI launch.
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors.
Understand AI from first principles.
• Deep Value Intelligence (multi-agent AI analyst): https://deepvalues.ai
• DeepValues for iOS (intrinsic value of any stock): https://apps.apple.com/us/app/deepvalues/id6775568608
Podcast & channels
• Spotify: https://open.spotify.com/show/2FlcJXyfpOMhbSFhxYRSqu
• YouTube (AI): https://youtube.com/@deep-values
• YouTube (investing): https://youtube.com/@TheDeepValues
The two hosts' voices are AI-generated from the sources discussed. Educational content only; not investment advice.
#AI #ArtificialIntelligence #MachineLearning #LLM #DeepValues
-
20 min
$220B AI Bet: How AWS Custom Chips Cut Costs 40%
Show notes
AWS plans to spend $220B on AI infrastructure in 2026 while its data centers are already sold out through 2027. We break down the capex cash cycle, why inference is still too expensive for most enterprises, and how Graviton and Trainium cut costs 30-40%.
Deep Values explains AI from first principles: the research papers, models, chips and companies shaping the AI era, and what they mean for builders and investors.
Understand AI from first principles.
• Deep Value Intelligence (multi-agent AI analyst): https://deepvalues.ai
• DeepValues for iOS (intrinsic value of any stock): https://apps.apple.com/us/app/deepvalues/id6775568608
Podcast & channels
• Spotify: https://open.spotify.com/show/2FlcJXyfpOMhbSFhxYRSqu
• YouTube (AI): https://youtube.com/@deep-values
• YouTube (investing): https://youtube.com/@TheDeepValues
The two hosts' voices are AI-generated from the sources discussed. Educational content only; not investment advice.
#AI #ArtificialIntelligence #MachineLearning #LLM #DeepValues
Technology lists
More Technology shows picked out from the charts
More in Technology
Placed highest on Apple Podcasts' Technology chart right now