Full Workshop: Setting Yourself Up for Success —Jason Liu, OpenAI Codex

By ai.engineer

Categories: AI, Tools

Summary

Jason Liu from OpenAI reveals how AI agents solved coding and operations work, shifting knowledge work complexity from doing 10 things to tracking 200 concurrent projects. He uses voice-input foot pedals, 5-week-old threads with 400+ sub-agents, and team plugins to become a force multiplier—proving AI productivity isn't about token maximization but strategic delegation.

Key Takeaways

  1. Implement 'compaction'—consolidate work into persistent threads that can maintain context across 5+ weeks with 400+ sub-agents, eliminating constant context switching and improving coordination.
  2. Switch from text input to voice commands using foot pedals (transcribe/enter buttons) to stay hands-free and context-focused on human collaboration rather than Slack monitoring.
  3. Build shareable team plugins and skills rather than hoarding AI productivity—the most popular skill (finalize codeex app) triggers before every pull request and creates company-wide multiplier effects.
  4. Scale by tracking personal memory through conversation threads instead of trying to remember projects—enables delegation visibility and reveals automation opportunities across 200+ concurrent initiatives.
  5. Optimize for token efficiency through strategic workflows rather than token maximization—focus on which documented patterns actually work to avoid wasting resources.

Related topics

Transcript Excerpt

[music] All right, let's just uh kick things off. How many people here already saw the keynote that I gave? Okay, not everyone. That's good. This talk is effectively going to be a stretched version of what I had given in the main stage except two things. One, I want to give you some time to try to set things up yourself, you know, Wi-Fi gods permitting. And then also two, be a little more interactive. I I had to be very high level when I was talking about what I use Codex for, but here if you have any questions, we have like 70 minutes. If you have any questions, just raise your hand and we can start answering some of these things, especially because not a lot of the workflows have been really well documented. And so if you if you're very curious on how things work, I'm really happy to uh …

More from ai.engineer