Skip to content
Matthew Blode
Guide

Chapter 6 of 6

How I actually use AI

The ten lines I actually run.

The tools keep changing. My loop doesn’t:

  1. Put the outcome and definition of done in the first prompt.
  2. Ask the agent to research the repo and propose a plan.
  3. Correct the plan before it writes code.
  4. Give independent work to parallel agents.
  5. Pick Claude Code or Codex for the job.
  6. Load a skill instead of retyping a checklist.
  7. Review the diff, then simplify it.
  8. Open the product and use it.
  9. Test according to what can break.
  10. Ship it, then watch what happens.

Turns out the code is the cheap part. Direction, taste, ownership, and talking to users still belong to me.

Read the long version: Method · Uses