Drew Breunig · 2026-05-04 · notable
Drew Breunig: 10 Lessons for Agentic Coding — What Should We Do When Code Is Cheap?
Drew Breunig distills a year of working with Codex, Claude Code, and Pi into 10 habits for teams whose code generation is suddenly cheap — and why maintenance, support, and security never were.
If your agents can write any code on demand, what habits actually matter? Drew Breunig's 10-item field guide.
What is it?
An essay on practicing agentic coding day-to-day, drawn from production use of Codex, Claude Code, and Pi. The thesis: code generation is now cheap, but maintenance, support, and security are not — agentic code is 'free as in puppies.'
How does it work?
The 10 lessons cover: implementation as a learning step, frequent rebuilds, comprehensive end-to-end tests, capturing decision rationale, keeping spec and code in sync, focusing human effort on the genuinely hard problems, automating routine tasks, cultivating taste, leveraging expertise to write better prompts, and treating maintenance/security as the persistent cost.
Why does it matter?
Most agentic-coding writing is either hype or skepticism. This is a working checklist from a practitioner, useful for teams adopting Codex/Claude Code/Pi without yet rewriting their habits — and for individual engineers trying to keep pace.
Who is it for?
Engineering managers and senior ICs adopting AI coding agents.