Built in 30 minutes. Now earning $50,000 a month, he says.
Pieter Levels’ own figure, reported by 404 Media, March 2025.
A course for people who code with AI
Seven short chapters. Each one teaches a habit, gives you real tasks, and shows you how to check your own work, so you can trust what the AI changes. For developers who feel behind, first-time builders with an idea, and teams that need everyone on the same page.
7 chapters · 28 lessons · 128 workshop tasks · no card needed
Real results, with sources
Built in 30 minutes. Now earning $50,000 a month, he says.
Pieter Levels’ own figure, reported by 404 Media, March 2025.
Rakuten ships a new feature in 5 days, not 24.
A cheese shop got $180,000 in shipping overcharges back.
No technical training. 45 days. 10,000 users.
Sabrine Matos, Plinq, in a Lovable customer story, September 2025.
01 · I write code
Hooks, skills, subagents, evals: the chapters go from the basics to running many agents on one job, each with tasks you check yourself.
See the chapters →02 · I have an idea
Start with what a terminal is and how to undo mistakes. By chapter 4 you write a one-page spec and ship a first working slice without a team.
Start chapter 1 →03 · I run a team
Live training for your engineers, recordings of every session, and office hours while they build the habits, scoped and priced before anything starts.
Ask about training →Free · no card needed
The whole first chapter, and four short reads from tests we ran ourselves.
Chapter 1 · Free
4 lessons · 23 workshop tasks
You want an AI assistant that works inside your own project, and you want to trust what it changes.
You may know Claude from the website or the app; this chapter moves you to Claude Code in the terminal, gently: what a terminal is, why every lesson here uses it, and how to undo any mistake. Then it teaches three habits: give Claude a memory, ask clearly, and check its work. The harder chapters build on them.
Setting up Claude Code
Why Claude Code belongs inside your project, and how to start safely
How Claude remembers
Why Claude forgets between sessions, and the two files that fix it
Prompting
How knowing prompting is key to everything else
Checking Claude's work
Why a check you can run beats a reply you read
For developers · Tested
Six small coding tasks, 100 runs: did Claude say "done" while the code was wrong?
Read →For developers · Tested
A check script Claude can run got 59 of 60 replies to follow three rules. The same script, never mentioned, got 0 of 60.
Read →For developers · Tested
The same three rules in CLAUDE.md files of 5, 61 and 201 lines, 180 replies: the short file was followed more often.
Read →For developers · Tested
A rule in CLAUDE.md held for 43 of 60 replies. With a Stop hook it held for 58 of 60, by a second grader.
Read →How a lesson works
Every lesson is the same shape, so you always know where you are. Nothing is graded for you. Every task ends with a check you can run and see.
Read
Why this matters and the one idea of the lesson, in a few minutes.
Workshop tasks
Four to seven short tasks, in a sandbox or on your own machine, each with a line that says what 'done' looks like.
Recap
The habit in three lines, so it sticks.
Takeaway
A worksheet, template or starter kit you keep and reuse on your real project.
Start free. Upgrade when you want every chapter. Teams and companies get live training.
$0
All you need to start building and automating, without guesswork.
$29a month
Every chapter and the forum.
By quote
A half-day of training for your team.
By quote
For the company that pays for AI, so every engineer uses it well.