What we learn making teams AI-first: methods, numbers, mistakes included.
Subscribe via RSSHow a fifteen-minute prompt put an untested project under the strictest quality control: a goal the agent checks itself, coverage first, a ratchet against shortcuts.
How to feed an AI reviewer, test it on past PRs, and decide when a human should look.
Agents write the code and humans review all of it. Five reasons, mostly psychological, that keep us from starting to delegate review.
Agents write the code, humans review all of it. An inefficient way to work and needless suffering, and four questions to get out of it.