The 8 Layers of Agentic Engineering
Most developers use AI the wrong way.
They prompt, wait, check the diff, then prompt again.
That is not Agentic Engineering.
Prompting is just step 1 of 8.
Here is the roadmap to become an AI Agent Engineer:
Fundamentals: prompt, context, harness, memory and loop engineering
Harness: steer agents with instruction files, skills, subagents and hooks
Verification: catch bad agent code with mutation, property and end-to-end tests
Review: review intent and the WHY instead of reading every line of code
Memory: create a centralized second-brain for all your agents
MCP: build your own MCP server to give other agents access to your product
Loops: don’t prompt your agents. Make your agents prompt themselves
AI employee: make your agents run 24/7 on your own sandboxed server
AI Agent Engineer launches next Monday
I am super excited to share that next Monday, 26 October at 08:00 CET, I launch my new course that teaches all 8 layers of Agentic Engineering.
To thank my readers, I start with a huge 33% discount, so you can save $100 in the first 7 days.
I went all in on AI coding in 2023. This year alone, I built hundreds of AI agents for companies.
My own coding agents run 24/7 on my servers and in the cloud. They write, test and ship code while I sleep. They also handle the admin and operations of my business.
You know me for Test-Driven Development and Extreme Programming. With AI agents, that craft matters more than ever.
Agents write code fast. But they also produce huge amounts of AI-generated code and open a lot of pull requests, which require new software disciplines and workflows.
In the course, I show you how to build, steer, verify and run AI coding agents for scale and production systems.
What you get
Here are all the things you get in the course:
✅ 3.5+ hours of hands-on video: 8 modules, 56 lessons
✅ 27 hands-on tasks on a real world app with Next.js and Postgres
✅ The Agentic Coding Handbook (90+ pages)
✅ The course kit: 10+ skills, hooks pack and AI employee config pack
✅ Completion certificate
✅ Lifetime access and every update
➕ Optional: private 1:1 mentoring
3.5+ hours of hands-on video
Most AI courses teach dry theory with little to no room for practice. I built my course around hands-on exercises, so you can learn by doing:
You build every layer yourself on an app written with TypeScript, Next.js and Postgres. The course contains 8 core modules, 56 lessons and 27 hands-on tasks, each with setup guides and solutions. More than 3.5 hours of video content.
The Agentic Coding Handbook (90+ pages)
A 90+ page reference for agentic engineering: prompts, context, harness, verification, review, memory, loops and security. It also covers material outside the lessons, like guardrails for agent code, test doubles, TDD with an agent, and when to use a CLI, a skill or an MCP server.
The Course Kit: skills, hooks and configs
A battle-tested library of 10+ skills, the hooks pack, the config pack for your 24/7 AI employee, and the second-brain template. Point your agent at a guide and say “set this up”.
Completion certificate
Finish the course and get a certificate of completion. Add it to your CV or LinkedIn, or use it to claim your company’s training budget.
Lifetime access and updates
Pay once, keep it forever. AI tools change every month, so I update the lessons and the kit when they do. You get every update for free, with a changelog.
Private 1:1 mentoring (optional)
Want my eyes on your setup? The coaching option adds a private 60-minute call with me and 1 week of direct WhatsApp support. We work on your codebase, your agents and your questions.
Is This Course For You?
The course is for engineers who already use an AI coding agent like Claude Code, Codex or Cursor. You want to trust AI-generated code without reading it, and you want agents that test their own work and ship to production autonomously.
If you want to ship 10× more code while reading 10× less of it, then this course is for you.
Who should skip it? People who are not into coding.
What Engineers Say about The Course
Huge Launch Discount
To thank you for reading this newsletter, I start the launch with a huge discount.
From Monday 26 October 08:00 CET until Monday 2 November, the course costs only $199 instead of $299 (33% OFF).
⚠️ Then the price goes up to $299.
Buying for your team? Teams of 3 or more get 25% off every seat: $149 instead of $199 during launch week. Email me, or reply to this email, and I send one invoice for the whole team.
Want personal help? Add the 1:1 coaching option: $999 during launch week, then $1,499.
🛡️ 14-day money-back guarantee: If the course is not for you, email me within 14 days. I refund you the same day, no questions asked.
Got questions?
Comment below, or hit reply and tell me in one sentence: what does your AI agent get wrong most often?
See you next Monday, my friend!
PS: Mark your calendar. AI Agent Engineer opens on Monday, 26 October at 08:00 CET, at $199 for the first 7 days.