Writing

  1. Gradual determinism via do-nothing scripts for agentic workflows
  2. Project-specific evidence can justify skipping review of AI-generated code
  3. Using screenshots as a specification for AI code regeneration
  4. Rules for using unreviewed AI-generated code
  5. Human code review is still cheaper than fully automated verification
  6. Toward automated verification of unreviewed AI-generated code
Diagram of a cyclical dependency and some question marks