I think assistants / agents are going to be the big thing this year. I was working on something at the end November that was proposing competent PRs based upon

I think assistants / agents are going to be the big thing this year. I was working on something at the end November that was proposing competent PRs based upon request for work in a GH issue. I was about halfway through the first iteration of a prompt role that can review, approve and merge these PRs. End goal being a fully autonomous software factory wherein the humans simply communicate via GH issue threads. Will probably be back on this project by mid February or so. Really looking forward to it. Bigger, more useful context is all I think I really want at this point. The other primitives can be built pretty quickly on top of next token prediction once you know the recipes.