Lex Fridman Podcast · 4 min instead of 3 h 16 m · Feb 12, 2026
Empathize with the agent's blank-slate view of your code
Creator, OpenClaw
Lex FridmanHost · Host, Lex Fridman Podcast
Robotics and AI researcher at MIT; started the show in 2018.
About this segment
Peter says people who fail with coding agents forget the agent starts every session knowing nothing. Guide it with pointers, watch for friction, and treat it like a capable engineer.
From “OpenClaw: The Viral AI Agent that Broke the Internet - Peter Steinberger | Lex Fridman Podcast #491”, Feb 12, 2026. Watch the full episode
Takeaways
- Agents start fresh each session, so point them where to look
- Slow runs signal missing context or an architecture mismatch
- Don't force your own worldview, such as renaming things they choose
In this episode
Full episode- 34:02Squatters sniped his accounts in seconds during the rename4:01
1:05:46Empathize with the agent's blank-slate view of your code3:47
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 1:05:59
that's sometimes the impression i get because it does not it needs a different level of thinking you have to learn the language of the agent a little bit understand where they are good and where they need help you have to almost consider consider how codex or claude sees your code base like they start a new session and they know nothing about your product project and your project might have hundred thousand of lines of code so you gotta help those agents a little bit and keep in mind the limitations that context size is an issue to like guide them a little bit as to where they should look that often does not require a whole lot of work but it's helpful to think a little bit about their perspective as weird as it sounds i mean it's not it's not alive or anything right but but they always start fresh i have i have the the system understanding
- 1:07:04
so with a few pointers i can immediately say hey i want to like make a change there you need to consider this this and this and then they will find a look at it and then their their view of the project is always it's not never full because the full thing does not fit in so you have to guide them a little bit where to look and also how you should approach the problem there's like little things that sometimes help like take your time that sounds stupid but and in 5.3 that was partially addressed but those also oppos sometimes they are trained um with being aware of the context window and the closer it gets the more they freak out literally like some sometimes you see the the real raw thinking stream what you see for example in codex is post-processed sometimes the actual raw thinking stream leaks in and it sounds like something like from the borg
- 1:08:02
like run to shell must comply but time and then they they like like that comes up a lot especially so so and that's that's a non-obvious thing that you just would never think of unless you actually just spend time working with those things and getting a feeling what works, what doesn't work. You know, like, just as I write code and I get into the flow and when my architecture's not right, I feel friction.
- 1:08:38
Well, I get the same if I prompt and something takes too long. Maybe, okay, where's the mistake? Do I have a mistake in my thinking? Is there, like, a misunderstanding in the architecture? like if something takes longer than it should you can just always like stop and like just press escape where are the problems maybe you did not sufficiently empathize with the perspective of the agent in that sense you didn't provide enough information and because of that it's thinking way too long yeah it just tries to force a feature in that your current architecture makes really hard um like you need to approach this more like a conversation for example when i