A team lead's guide to integrating AI coding tools into team workflows, focusing on the 'amplification principle': AI magnifies existing team practices, good or bad. Covers what code AI should and shouldn't write, a decision tree for AI generation, specific categories of technical debt AI quietly introduces (initialization, load transition, security surface, operational debt), a metrics framework for measuring sustainable velocity gains, practical process templates (plan.md, PR review standards, commit message formats), and six unsolved structural problems in AI-assisted development including hallucination, context attenuation, and true cost accounting.
Table of contents
The Amplification PrincipleWhat AI Should and Shouldn’t WriteMeasuring What Actually MattersThe Technical Debt AI Quietly CreatesHow Teams Actually Ship With AITool Selection: Evaluate, Don’t MarryThe Unsolved ProblemsThe Bottom LineSort: