Using LLMs in the job
Table of Contents
We decided that it would be better for the company to increase the use of LLMs for implementing, planning and reviewing the PRs. This decision hurts me a little, I dislike how this tech requires using SaaS from a bigtech, but during the last 6 days () working with it, the productivity increased for sure.
The flow focus point we reach while in a coding session is not really reachable, and the anxiety is increased, but the output gets nicer if we iterate the subagents: master<->(developer<->reviewer).
I've been using opencode in a Claude user plan. Since is a thirdy party app, it got rate limit. So I'm now with the same agents/skills setup in openclaude, which also works well.
End of the history
Fortunately I have my job things isolated in a qube, so the personal things remain how I want: Privacy is requirement.