Are you an AI maximalist?
No.
Setting aside the conversation about “the AI”, which is everyhere, and focusing on a domain where the new tools have an evident impact: coding.
Most people’s experience of “AI for coding” is copilot doing a PR review, or ChatGPT, best case they have Claude via API in their IDE.
When you try to describe something different, you sound like you’re selling something. Or worse, like a true believer.
Come on too strong and you’re an AI maximalist.
Hard to find a middle ground when you don’t share the same reference points, and most people operate with outdated information about what “AI coding” means.
But wander too much on the other side, and suddenly people tell you “it thinks” or “it makes intelligent decisions”…
All I see is better training data, better context handling.
How you use the tool is somewhere on a spectrum: from not using it to full on Gas Town where you orchestrate teams of teams of agents.
In the middle of the spectrum, there are many high value, low risk use cases.
You’ll find some examples here, setting up a review board, writing docs, or chores (checking for misalignment between the doc and the code), …
You’ll also find some experiments which are more on the far side of the spectrum.