This week's episode is with Ed Bayes and we're doing a deep dive into how designers at OpenAI prototype the future of Codex/ChatGPT Work.

Some highlights:

- How prototyping at OpenAI has shifted recently

- The design challenges of bringing Codex into ChatGPT

- How Ed thinks about the spectrum of dynamic interfaces

- How Codex thought about their information architecture bets

- Why Ed structures design teams around personas not features

- The traits of the designers making the biggest impact at OpenAI

- + a lot more

The Bitter Lesson (http://www.incompleteideas.net/IncIdeas/BitterLesson.html) — ML essay he referenced on scaffolding and model releases