[muratsu]

I find Codex and Claude Code to have different strength/weaknesses and wanted to be able to use them from a single interface. Currently hacking on https://devfleet.ai to make agent management more easy on myself.

Briefly mentioned on the article but async agents really thrive on small and scoped issues. Imagine hooking them up to your feedback tool (eg canny) and automatically having a PR as you review the customer feedback. Now this would likely not work for large asks but for smaller asks, you can just accept the PR and ship it really fast!

[conradkay]

Cool project! Do you think a lot of Codex's strengths are just from using GPT-5 as the model?

[muratsu]

The codex model is trained differently than the normal models. It has extra training on how to use cli and I find it to be better at project scope tasks (eg running tests, migrations, etc). Whereas in my experience Claude is the better coding model.