
We help founders make something people want. The Y Combinator Podcast is where builders talk about building. From the earliest days of an idea to scaling a company that changes the world, YC partners and founders share real stories, lessons, and tactics from the frontlines.

This episode shows what changes when AI stops acting like a chat tool and starts acting like a coworker with tools, identity, and an ongoing job. For PMs, the key lesson is that faster execution raises the value of clear objectives, careful permissions, and stronger product judgment.

Replit built its coding agent after watching people who had never coded paste errors back and forth between ChatGPT and its editor. Amjad Masad explains why enterprises now swap $10 million software projects for $250,000 of Replit, and why rationing AI tokens misses the point.

Francois Chaubard spent three months training a people-tracking model that retailers asked for and then refused to pay for. He explains the research habits that quietly kill technical products, and why a human in the loop should come before any model.

Supabase is seeing roughly 70% of new activity come from agents and more than a million new databases created each week. Paul Copplestone explains why that changes product design, why personal software may explode, and why the real bottleneck is shifting from writing code to knowing what to build.

Marvin Schwaibold explains why AI is moving design work out of static mockups and into fast, working prototypes. For PMs, the real lesson is not the tool list. It is that when output gets cheap, judgment, clarity, and product taste become the bottlenecks.

Farza Majeed argues that launch videos should be a constant operating rhythm, not a rare polished event. The useful lesson for PMs is not filmmaking technique alone, but a better product loop: work backward from a demo, tell a story, and judge every channel by conversion instead of views.

Magic Patterns CEO Alex Danilowicz argues that AI has made building much cheaper, so the real product work has moved upstream into clearer exploration, better context, and higher-converting interfaces. The episode is especially useful for its teardown of common "vibe coded" website mistakes, from hidden CTAs to fake inputs and overdone motion.

The founders of Waddle Labs and RoboCurve explain why coding agents are turning into robot controllers, why latency is now the real blocker to real-time robots, and why they think general-purpose robots are about two years away.

YC's partners share the batch data behind three shifts nobody predicted at this speed: startups moving from bits back to atoms, solo founders nearly quadrupling, and companies hitting real revenue in weeks instead of years.

A YC visiting partner who closed millions through founder-led sales breaks down why targeting, not polished copy, is what actually moves reply rates.

Ollama's CEO, who watches token flow from 9 million developers and 85% of the Fortune 500, explains why open models went from hobbyist curiosity to enterprise default in under two years, and why the real bottleneck has shifted from model intelligence to what runs on top of the tokens.

Paul Graham explains why the founders who build billion-dollar companies are driven day to day by fear of disaster rather than fantasies of wealth, why AI arrived backwards from how researchers expected, and why "starting lean" is still the right instinct even in an era of expensive AI compute.

Heart Aerospace went from a 3D-printed model you could hold at YC Demo Day to the largest electric airplane ever flown, in seven years. Founder Anders Forslund on picking the segment incumbents ignore, tying every fundraise to something physical you can touch, why looking conventional is a feature, and the SpaceX risk philosophy that reshaped the whole product.

Three non-lawyers in a windowless Stockholm conference room took Legora from $1M to $100M ARR in 18 months. Max Junestrand on why they refused to fine-tune models, why they froze all sales for six months, and why the ability to evaluate models is the real IP.

The designer who drew the Happy Mac, the command key, and the Chicago typeface on 16x16 grids explains why less detail makes an icon more universal, why a metaphor outlasts a literal product drawing, and the design and career advice she has collected over a lifetime.

Chelsea Finn on why physical AI has almost no room for the "mostly right" mistakes consumer AI gets away with, the reinforcement learning recipe that pushed a robot to over 90% reliability making espresso, and why she thinks robotics just had its own GPT moment.
Everyone hears the same episodes.
Few can do what they describe.