The Two-Tool Stack That Stuck

I tried the 'one AI to rule them all' approach and the 'twelve specialized tools' approach. The one that stuck was two, a careful model for writing I care about and a fast one for everything else.

The careful one thinks, holds a long thread, and pushes back when I'm wrong. The fast one answers, summarizes, and gets out of the way. Neither is 'the best,' they're the right tool for two different jobs.

Adding a third always felt productive and always died within a week, because the switching cost exceeded the gain. The two-tool stack survived because the boundary was obvious: does this matter, or just needs doing.

If your stack keeps growing, you're collecting potential, not using tools. Cut to the two you touch daily and put the rest on a 'miss it to resubscribe' challenge. Most you won't.