~/workshop $ run agentic-engineering --mode workshop
Agentic Engineering Workshop
bettervibe.
Avoiding the vibe coding hangover.

Vibe coding gets features shipped fast. Agentic engineering gets them to production. This workshop teaches the discipline: directing AI agents as force multipliers while keeping your engineering judgment in charge. Less prompting into the void — more architecting, reviewing, and owning the output.

Register → See the curriculum
Date
May 13, 2026
Time
1:00 – 6:00 PM
Location
WERK1, Munich
02 — The problem

Vibe coding has a ceiling.

Prompting your way to a prototype is easy. But vibe coding has failure modes that only surface after you've shipped — and they compound fast in production.

// pain point 01

The review bottleneck

AI writes pull requests faster than humans can review them. Nobody wants to read AI-generated code all day — the ownership gets blurry, quality slips through, and the review process becomes a choke point nobody wants to touch.

// pain point 02

The output you can't control

Same prompt, different result. The agent drifts, makes architectural choices you didn't ask for, and builds things you didn't want. Getting consistent, predictable output from AI coding tools is harder than it looks.

// pain point 03

The context collapse

Long sessions degrade. The model forgets what it decided two hours ago, contradicts its own earlier work, and starts producing noise instead of signal. There's no memory — only whatever fits in the window right now.

// pain point 04

The vibe coding hangover

The feature shipped. The code works — until it doesn't. You own something you don't fully understand, written faster than you could reason about it. Tech debt compounds. And refactoring AI-generated code is worse than writing it from scratch.

03 — Who it's for

This workshop is for you if—

✓ You should come

  • You write code daily — professionally or on side projects
  • You've used Claude Code, Cursor, or a similar AI coding tool
  • You can read a diff and know when something looks wrong
  • You're a solopreneur, freelancer, or early-stage founder who ships code
  • You want concrete techniques you can apply the very next day
  • You've vibe coded your way into tech debt and want the discipline to back it up

✕ This isn't for you if—

  • You've never written or read code before
  • You're looking for a no-code / Lovable-style workflow
  • You want theory without hands-on practice
  • You expect finished answers rather than practiced techniques
04 — Curriculum

What you'll learn.

Concrete agentic engineering techniques — from structuring your agent's context to knowing when to take the wheel back. Built for developers who want to move beyond vibe coding and ship code that holds up.

Foundation

Structuring for the agent

What goes in CLAUDE.md — and what doesn't

  • What makes a good CLAUDE.md / agents.md
  • Skills: what they are and when to use them
  • What to put where: context file vs README vs commands
  • Static code analysis tools and why they matter more now
  • Agent reviews: using external tools like CodeRabbit
Core concept

Agentic engineering

You set the direction. Agents execute. You own the result.

  • Calibrating your level of control: too much vs too little
  • How to drive the model to get the highest quality output
  • Playing to Claude Code's strengths, avoiding its weaknesses
  • Breaking work into chunks so the agent doesn't blow up its context
  • Where to draw the line: what you delegate vs what you own
Quality

Keeping the code yours

Sustainable, defensible, production-ready output

  • How to write clean code with AI — not despite it
  • Test loops: building confidence without constant manual testing
  • Git branches and worktrees for AI-assisted workflows
  • Code quality metrics and how to track them
05 — Format

How the afternoon runs.

1:00 PM

Introduction (60 min)

Shared foundations before we get hands-on. CLAUDE.md best practices, compound engineering, context window fundamentals, and how to structure your workflow for the session ahead.

2:00 PM

Project time (2 hrs)

Hands-on guided work with instructor support throughout. You'll apply the techniques from the intro directly — with real code, real problems, and real feedback.

4:00 PM

Networking + beers

The workshop ends, the conversation continues. Beers, debrief, good conversations.

06 — Instructors

Who's teaching.

Not academics — practitioners. All three ship production code with AI daily.

Traun Leyden

Traun Leyden

Cofounder of Fluensy.app - Speak as brilliantly as you think. Ex-Databricks / Couchbase senior engineer with 20+ years shipping backend systems in Go, Python, and Rust. Traun brings deep experience in building reliable, well-tested software — and knows exactly where AI-generated code breaks down in production.

Adam Charnock

Adam Charnock

Founder of Lithus, where he manages bare-metal Kubernetes infrastructure for SMEs. Previously contracted at Twitter and Blocknative. Adam has run production systems at scale for 18 years — he teaches how to calibrate control and let agents handle the right things.

Dominik Grusemann

Dominik Grusemann

Co-Founder of Marbles AI and former CTO of Chatchamp (acquired 2023). Previously led agile engineering at BMW's autonomous driving division. Dominik focuses on CLAUDE.md best practices and compound engineering.

07 — Pricing

Reserve your spot.

~/events/may-2026 $ ls -la
Date May 13, 2026 — Wednesday
Time 1:00 PM – 6:00 PM CEST
Venue WERK1, Munich
Format Hands-on · Projects provided
Refunds Up to 7 days before the event
API keys Provided — no extra cost to attend

Bootstrap / Startup

Solopreneurs and early-stage founders

€75

Early Bird

Individual developers

€125

Regular

Standard ticket

€250
Register on Luma →
08 — FAQ

Quick answers.

Do I need to bring my own project?
No — we'll provide a selection of projects to choose from. You just need your laptop and your development environment set up. Setup instructions will arrive by email before the event.
What if I'm new to Claude Code?
Comfort reading and writing code is the baseline. The workshop assumes you've used AI coding tools, but we'll cover the fundamentals in the intro. You don't need to be an expert.
Is there a refund policy?
Full refund up to 7 days before the event. After that, your spot can be transferred to someone else.
Can I expense this through my company?
Many companies have training budgets that cover this. Reach out at dominik.grusemann@gmail.com and we can help figure it out.
Which tools do I need installed?
We recommend Claude Code or OpenAI Codex. We'll provide API keys so you don't need to worry about cost. Cursor users are welcome but won't receive dedicated support.
Is there food and drink?
Water and sparkling water during the workshop. From 4 PM: beers, soft drinks, and snacks. You won't go hungry.
09 — Updates

Stay in the loop.

Get notified about future dates, early-bird pricing, and new workshop topics.

10 — Sponsors

Backed by builders.

Companies that share our belief in shipping AI-assisted code that holds up.

~/workshop $ register --now

Pick a project.
Leave with techniques.

A hands-on afternoon for developers who are done vibe coding and ready for the real discipline. Leave with agentic engineering techniques you can apply the very next day.

Register → Any questions? Write us

May 13, 2026 · WERK1 Munich · 1–6 PM