I have deliberately chosen to keep the way I interact with large language models extremely simple, even primitive. I don't get all tangled up in spawning subagents, accumulating skills, or trying to build a software factory that runs while I'm not there. I have conversations with my robot friend in which I ask it to analyze and generate code, and that's it.

I have yet to see any convincing evidence that this is losing me productivity.

The reason I choose to keep it simple is because it seems to me that all these elaborations are ways to cope with the frontier models not yet being quite smart enough. But the models are always getting smarter.

If both of those propositions are true, why clutter my brain with procedures that are likely to have only a short half-life of usefulness? I prefer to spend my think time concentrating on the actual problems.

Yes, I'm applying the bitter lesson to my own workflow. AI isn't powerful enough yet? Do the Zen thing; wait, and be aware. It's not like I'm short of problems for which LLMs are already massive help.