korrents
Ed Bayes

What Ed Bayes thinks about LLMs

@ed-bayes · 18 positions · 0 changes of mind

Member of Design Staff at OpenAI, where he works on coding agents and leads design for Codex and ChatGPT Work.

Everything they publish, on ppll ↗

Ed Bayes did not write this page.

We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.

5 dated positions, 2026, in their own words. Our reading of what Ed Bayes has said — not written or endorsed by them.

  1. Our readingA good framework for designing products is to design for six months out: presume model capabilities will get better and build for the world you would then be in.

    design

    Their wordsOne of the the key kind of frameworks I use here is like designing for 6 months out. So if you just presume that model capabilities are going to get better and you think about where we might be in six months and you kind of start to build products for that world that's kind of a good framework to to use

    ↗Ed Bayes - Designing 6 months into the future at OpenAIyoutube.com 1st of 18 in this recording

  3. Our readingWith model-generated interfaces, instead of specifying the pixels you move to the skill level and define a design language within it.

    design

    Their wordsinstead of like specking out the pixels, you basically move to the skill level and you kind of define a uh design language, you know, within the skill.

    ↗Ed Bayes - Designing 6 months into the future at OpenAIyoutube.com 7th of 18 in this recording

  5. Our readingScaffolding built around a model will get washed away by new model releases, so you have to stay flexible.

    Their wordsany kind of like scaffolding that you do like over scaffolding like some of the things that I just talked about will kind of get washed away with like new model releases. So, I think you just have to stay super flexible, right?

    ↗Ed Bayes - Designing 6 months into the future at OpenAIyoutube.com 9th of 18 in this recording

  7. Our readingThe medium of designing with models is nondeterministic enough that without real prompts flowing through it, an idea can look great in a design review and be completely different when users use it.

    design

    Their wordsthe medium that we're working with is like so nondeterministic that without the kind of water flowing through the pipes, it's it's hard to reason about, right? Like, you know, some of these ideas like sound great on paper, but when you actually like send a bunch of prompts, as I was saying, right, with the, you know, with the visualiz, you know, building like kind of golden evals, it's like it can look great in a design review or in a prototype, but actually when users use it, it's completely different.

    ↗Ed Bayes - Designing 6 months into the future at OpenAIyoutube.com 10th of 18 in this recording

  9. Our readingOne of the biggest barriers to using tools like Codex is assuming it can't do something; pushing past that keeps expanding the scope of what is possible.

    Their wordsOne of the biggest barriers of of of folks like using tools like Codex is this like limiting factor of like, oh, it can't do that. But every time, at least I've kind of like pushed myself, you know, really expands the scope of what is possible.

    ↗Ed Bayes - Designing 6 months into the future at OpenAIyoutube.com 18th of 18 in this recording