On AI

A developer shares their personal experience using Claude Code heavily in their workflow, focusing on what makes AI coding tools effective or frustrating. Key insights include: the importance of enjoying code reviews to get value from AI tools, using incremental review strategies for large changes, yak-shaving and nitpicking AI-generated code freely, and keeping sessions simple. Criticisms include inconsistent behavior, overly verbose comments, and false AUP flags. The author pushes back on code quality concerns about AI-generated code, arguing that poorly designed human code isn't inherently better. Closes with a concern about AI tool cost inequality in lower-salary markets.