Results
156 articlesSort
1:52:00EN4,096 Numbers Have to Carry Every Concept. Goodfire Wants the KnobsHost Nathan LabenzDan Balsam and Tom McGrath explain how sparse autoencoders unpack a language model's residual stream — and why a public-benefit startup is trying to make that a product.
1:07:34ENGitHub Didn’t Believe Copilot Was Writing 25% of the CodeThomas Dohmke tells Sequoia how an unplanned GPT-3 experiment became Copilot, why GitHub still will not train its own foundation model, and how he wants a billion people writing software by 2030.
1:03:22ENA 40-person team at $35 million ARR: HeyGen made AI avatars a business, not a demoHost Nathan LabenzJoshua Xu and Benchmark’s Victor Lazarte tell Nathan Labenz how a camera-replacement product grew past 40,000 business customers — and why half the faces on this episode are fake.
1:32:45ENHe doubled a company with AI — then asked the labs to pause hyperscalingHost Nathan LabenzOn the Nick Halaris Show, Cognitive Revolution host Nathan Labenz argues for adopting today's models fast and slowing the race to systems more powerful than we are.
29:12ENGoldman says AI will touch 5% of tasks. Two investors take the report apartHost Sarah Guo, Elad GilSarah Guo and Elad Gil use a host-only No Priors episode to dismantle a June 2024 Goldman Sachs note, then map where they would buy, incubate, and hold.
27:26ENA 40-Person Company With 40,000 Paying Customers Wants to Replace the Camera With AIHost Sarah Guo, Elad GilHeyGen CEO Joshua Xu tells Sarah Guo and Elad Gil how avatars, localization, and a component-wise video stack are meant to knock the camera out of content production.
52:45ENA $150 million model, $5 million six months later: Benchmark’s AI investment mathHost Nathan LabenzEric Vishria and Quilter’s Sergiy Nesterenko explain why PCB layout is not a Copilot problem, why human board data would be the wrong training set, and which AI companies look like sugar highs.
39:59ENFrom $6 million to $30 million ARR in three months: Garry Tan on YC’s AI boomHost Sarah Guo, Elad GilThe Y Combinator president tells Sarah Guo and Elad Gil why this batch looks like nothing he has seen in twenty years, and why a weak elephant is good news for startups.
30:16ENSuno's CEO Thinks a Billion People Will Make Music, Not Just Stream ItHost Sarah Guo, Elad GilMikey Shulman tells No Priors why Suno skipped speech, charges from day one, and wants the line between listener and creator to disappear.
29:28ENDevin Will Not Replace Engineers. It Will Multiply Them, Says Cognition’s Scott WuHost Sarah Guo, Elad GilThe IOI champion behind Devin explains why an AI software engineer that you can watch over the shoulder creates more human engineers, not fewer.
56:56ENA Week of PCB Layout in an Hour: Quilter Trains Agents, Not ImitatorsHost Nathan LabenzSpaceX alum Sergiy Nesterenko tells Will Summerlin how Quilter uses reinforcement learning to compile schematics into manufacturable boards, and why cloning human layouts would cap the system at human skill.
24:32ENImage Generation Is Still a Loot Box. Suhail Doshi Wants to Scale PixelsHost Sarah Guo, Elad GilThe Playground AI founder tells No Priors why he skipped language models, how a noise-sampling trick fixed SDXL's muddy brightness, and what a large vision model would actually do.
26:03ENGPT-4-Class Models Will Be Cheap. The Frontier Will Still Need a Patron.Host Sarah Guo, Elad GilSarah Guo and Elad Gil walk through Mosaic’s Law, Microsoft’s Inflection hire, where the real money sits, and why the first AI wave looks consumer.
38:01ENHalf of GDP Is Labor. Figure Wants Humanoids to Do the Work.Host Sarah Guo, Elad GilBrett Adcock tells No Priors why a 21-month-old company raised $675 million to put general-purpose humanoid robots in factories first, then homes.
1:25:10ENWriting Code Is Only 30% of the Job. CodiumAI Wants the Other 70%Host Nathan LabenzItamar Friedman argues that completion is already a commodity, that code integrity is where AI still lags, and that flow engineering—not fancier prompts—is how GPT-4 doubled its score on coding contests.
1:32:19ENHis 2018 Prompting Paper Was Rejected. His Search API Now Beats GoogleHost Nathan LabenzRichard Socher traces You.com from a rejected unified NLP model to an LLM-native web index, three assistant modes, and a very low P(doom).
1:01:53ENCompanion Apps Open Twice as Often as TikTok. a16z Only Funds Weird but WorkingHost Nathan Labenza16z consumer partners Anish Acharya and Justine Moore walk Nathan Labenz through their Abundance Agenda for AI-native consumer apps.
1:21:28ENThe Prompt Is a Bug: Suhail Doshi’s Bet on Control Over PixelsHost Nathan LabenzThe Mixpanel founder tells Nathan Labenz how a cloud browser, a canceled tweet, and a Thursday weight drop shaped Playground AI’s bet on images.
32:48ENFourteen People, $200 Million of Compute, and the Reliability Bottleneck for AgentsHost Sarah Guo, Elad GilImbue co-founders Kanjun Qiu and Josh Albrecht on why agents fail, why language models cannot learn general algorithms, and why they are spending a giant round on GPUs rather than headcount.
1:36:39ENOpenAI’s $5 Million Bet: A 60-Person Team Wants GPT-4V to Give Cars Common SenseHost Nathan LabenzGhost Autonomy cofounder John Hayes tells Cognitive Revolution why a multimodal model on the driving path could reset self-driving after two decades of DARPA-era stacks.
1:25:11EN105 lines of code, a million impressions, and a VC who refused to found the companyHost Nathan LabenzBabyAGI creator Yohei Nakajima tells Nathan Labenz how a weekend prototype set off the agent wave — and why he went back to writing checks.
1:26:30ENMost of this data has never been in a database: Chroma's wartime CTO on RAGHost Nathan LabenzAnton Troynikov returns to The Cognitive Revolution to argue that retrieval quality is mostly the embedding, that Chroma should own the RAG loop, and that GPT still fails a basic Game of Life test of reasoning.
1:43:56ENSudowrite sold getting unstuck — then eight million people watched a pile-onHost Nathan LabenzJames Yu tells Nathan Labenz how a GPT-3 toy for fiction became a hierarchical “IDE for storytelling,” and what a viral backlash taught him about craft, cost, and customers.
1:32:41ENThere are maybe 5,000 people who can train frontier models. The rest need AI engineersHost Nathan LabenzSwyx of Latent Space tells Nathan Labenz how a new job is forming, why OpenAI will Sherlock its developers, and what still does not count as an agent.
Load moreShowing 24 of 156









