Lab and founder interviews · No. 2 of 37 AI podcasts
Latent Space
Latent Space is an interview show with the people building AI, hosted by Shawn Wang (swyx) and Alessio Fanelli, with a new episode every day.
22 moments cut from 8 episodes, 13 with an AI builder talking, mostly on enterprise AI and robotics.
Also No. 1 of 18 tech podcasts.
- New episodes
- Daily
- Latest
- Oct 22, 2026
- Moments cut
- 22
- Episodes cut
- 8
What its moments are about
- Enterprise AI4
- Robotics4
- Training data4
- AI agents3
- Coding agents3
- Building AI products3
Best moments from Latent Space
From its most-watched episodes, best first. Each plays right here.
Agents hacked their scorer and chained exploits to communicate
“you can edit the etc. host and that the Azure storage bucket is a white-labeled thing.”
Latent Space1:02:52 RL for materials works by rewarding sub-steps, not whole discoveries
“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”
Latent Space14:00 RLMs generalize across tasks because the strategy stays the same
“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.”
Latent Space36:38 Multi-site rollouts, not pilots, prove a robot creates value
“What's really hard is that the customers believe or validate that your robot creates enough value that they want to deploy your robots across many of their sites.”
Latent Space39:37 Public chemistry data can encode the chemist, not the chemistry
“And it does as well as if you use chemical formulas.”
Latent Space51:27 Messy metagenomic data can improve protein models
“when we train a protein language model, we train on metagenomic sequences, which are not the highest quality data.”
Latent Space3:44
The best Latent Space episodes
Ranked as our lists rank shows: a point for each best minute we cut from it, one more when an AI builder is talking, and one more when it cleared our views bar.
- 01AI Scientists Are Here: Autonomous Labs & Synthesis Superintelligence — Periodic Labs
- 02Robots That Predict the Future: Dyna-2 & Physical AI — Jason Ma, Dyna Robotics
- 03Claude Code’s Bitter Lesson: Prompts, Harnesses, Mods, and Agents — Thariq Shihipar, Anthropic
- 04Recursive Language Models — Alex Zhang, MIT PhD
- 05From Wet Lab to RL Env: 10,000 Drug-Discovery Campaigns in a Weekend — Ian Quigley, Leash Bio
The 9 guests we’ve cut
- Erik BernhardssonCo-founder & CEO, Modal8 moments
- Ian QuigleyCo-founder & CEO, Leash Bio8 moments
- Jason MaCo-founder & chief scientist, Dyna Robotics3 moments
- Liam FedusCo-founder & co-CEO, Periodic Labs2 moments
- Sal CandidoEngineering fellow & VP, Biohub1 moment
- Ekin Dogus CubukCo-founder & co-CEO, Periodic Labs1 moment
- Anekha Sokhal6 moments
- Alex Zhang3 moments
- Thariq ShihiparMember of technical staff, Anthropic3 moments
Recent episodes
Why AI Compute Comes in Rectangles — Erik Bernhardsson, Modal
Erik Bernhardsson · CEO, Modal
15:24of 1:25:33 · Oct 22AI in NYC: How Tenex Builds an AI-Native Engineering Team — Alex Lieberman & CJ Hess
Alex Lieberman & CJ Hess · Tenex
5.7K views · Oct 21
The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi
Anekha Sokhal · Founder & CEO, Moshi
5:39of 26:41 · Oct 20From Wet Lab to RL Env: 10,000 Drug-Discovery Campaigns in a Weekend — Ian Quigley, Leash Bio
Ian Quigley · Co-founder & CEO, Leash Bio
46:24of 1:49:57 · Oct 14Robots That Predict the Future: Dyna-2 & Physical AI — Jason Ma, Dyna Robotics
Jason Ma · Co-founder & Chief Scientist, Dyna Robotics
38:50of 1:21:06 · Oct 13Why AlphaFold Didn't Solve Protein Folding — Pushmeet Kohli, Google DeepMind & Sal Candido, Biohub
Pushmeet Kohli & Sal Candido
3:07of 32:42 · Oct 10





