Damnit, I just switched to Claude Code from Codex not two months ago, and now I'm going to have to find time to try this. For my purposes (and for context I am a not developer but a former PM using LLMs to build internal tools for my business), SOTA models all seem to be past the threshold of coding skill necessary, and now the harnesses are becoming the more important differentiating factor. Glad to see some healthy competition there, and specifically glad to see you've got automations in there, as I've specifically felt the lack of those in CC.
Very interesting! I've been using Cursor for a while for my side project, but you inspired me to hook up my OpenAI API and test it!
Damnit, I just switched to Claude Code from Codex not two months ago, and now I'm going to have to find time to try this. For my purposes (and for context I am a not developer but a former PM using LLMs to build internal tools for my business), SOTA models all seem to be past the threshold of coding skill necessary, and now the harnesses are becoming the more important differentiating factor. Glad to see some healthy competition there, and specifically glad to see you've got automations in there, as I've specifically felt the lack of those in CC.
Daniel: *Finally jumps aboard the chugging Claude Code train.*
Anthropic: How about that, but in a neat Cowork app?
Daniel: Yes please!
Anthropic: It's MacOS
Daniel: Oh...
OpenAI: Just wait till you see our new Codex app!
Daniel: YES PLEASE!
OpenAI: It's MacOS
Daniel: I'm starting to see a pattern here.