Learn

This is everything I teach, free, with no signup and no email gate. I build a Mac app called Ducktate, and these are the things I get asked about most: what to check before you launch an app you built with an AI coding tool, how to learn GTM Engineering from scratch, how I run a team of AI coding agents, how I stop staring at a blank page, and a setup prompt that cuts the tokens an agent spends on finding code.

Each course is short, in order, and made of real steps you can do today. Start anywhere. Nothing here asks you for a card.

• 

  The Librarian

  Set up a private company brain with reusable setup, contributor, and scheduling prompts. Use local commands, librarian rules, and executive reports with an audit trail.

  • 1 prompt guide
  • Updated 2026-10-11
  Set up your librarian
• 

  The Company Brain Course

  Build a private company brain on GitHub. Your agents read it before they work, leave a note after, and one librarian keeps every page true.

  • 7 modules
  • 53 min total
  • Updated 2026-10-19
  Start the course
• 

  Launch Readiness

  Built an app with Claude Code, Cursor, Lovable or Bolt? Check data security, privacy, tracking, payments, email and AI features before launch. Not legal advice.

  • 11 of 11 stages ready
  • 193 min total
  • Updated 2026-10-09
  Start the course
• 

  GTM Engineering

  Learn go-to-market engineering from zero: twelve stages, twelve build projects, quizzes, flashcards and a resource library.

  • 12 stages
  • 261 min total
  • Updated 2026-10-07
  Start the course
• 

  The Orchestrator Course

  Run a team of AI coding agents: one brain, many cheap hands, and a bar you can measure.

  • 7 modules
  • 59 min total
  • Updated 2026-09-10
  Start the course
• 

  Talk to Think

  Replace the blank page with a talk-first process: explain it, dump it fast, then shape it.

  • 5 lessons
  • Updated 2026-08-26
  Start the course
• 

  Find Code With Graphify and Jev

  One copy-paste prompt gives your AI coding agent a local code graph and a cheap ranking model. Measured on 30 questions: 19% fewer tokens, same accuracy.

  • 1 prompt
  • Updated 2026-09-20
  Get the prompt