Go Slow to Go Fast with AI: Writing Better Code, Not More Code

Most developers using AI are optimizing for the wrong thing. Speed metrics and line counts are not quality — and AI makes that confusion worse, not better.

This video is a reaction to Nolan Lawson’s article “Using AI to write better code more slowly,” published May 2026: https://nolanlawson.com/2026/05/25/using-ai-to-write-better-code-more-slowly/

Topics covered:

• Why measuring AI productivity in lines of code or tokens used is misleading
• The multi-model PR review technique (Claude, Codex, Cursor Bugbot) — and why good prompts can replace it
• How working in small, self-contained modules reduces hallucinations and improves output quality
• Why the right approach to AI-assisted development is just good engineering: incremental, focused, iterative
• The “token maxxing” management trap — and why it’s as silly as rewarding bug counts
• The difference between bug-fixing work and feature development when using AI agents

Going slow with AI isn’t a limitation — it’s the point.