- Duration
- One day
- Participants
- 5 to 15
- Location
- At our office or yours
- Language
- Dutch or English
What to expect
During this training day you work in a real codebase, together with an AI agent on your own laptop. The format follows the idea of a kata: through practice and repetition, a new way of working becomes a habit. Every step builds on the one before, and every exercise ends with proof that your solution really works.
What you can expect:
The course and the practice project come as one package. You read the explanation in the browser and apply it straight away in the code beside it. After the training, it is all yours to keep.
You can work through the steps again later, at your own pace. Or you use the course as a reference whenever a question comes up in your daily work.
The full course is available in both languages, so you can keep learning on your own or bring colleagues along.
Our trainers are certified experts with extensive hands-on experience. They apply this way of working every day in client projects themselves.
How your day runs
The day has four steps that build on each other: from how an AI agent works to what changes in your role and in your team.
Start here
We start with how the day works and a few house rules: let the agent do the work, give it one task at a time, and always ask for the real result. The agent then installs the practice project on your laptop itself, and the first exercise gets you a first result straight away.
How this kata works · A first exercise with proof
Context, model, mechanisms
To steer an agent well, you need to understand what happens behind the scenes. You learn what an agent actually gets to see, why every request costs money and how to keep that down, and how to spot an answer that sounds convincing but is wrong.
Tokens and generation · Prompt · The agentic loop and tools · Context · Session · Agent patterns · Model, reasoning level and usage · Grounding and hallucinations · What you take away
Agentic engineering
The heart of the training. You learn to set up a project so the agent gets the right instructions with every task, and how automatic checks keep watch over its work while it runs. With loop engineering you give the agent a goal it can prove on its own: it keeps working and adjusting until an automatic test passes. You also practise steering when things go wrong, matching the approach to the task, and knowing when it pays to run several agents at once.
Setup · Setup demo · Domain language · Quality gates · Steering · Skills and scripts · Processes · Prove your work (loop engineering) · Parallel work · What you take away
Soft skills
The last step is about people. With an AI agent, the developer's role shifts: less writing code by hand, more deciding what should be built and checking that it is right. You learn to manage expectations, because a working prototype in a few hours is not yet a finished product, and to bring colleagues and managers along with the change.
Team and process · Expectations · Confidence
What you take away
Skills you put to work in your own job the day after the training.
Learning by doing
No dry theory: you get to work straight away on your own laptop. You let the agent do the work, and in every exercise you check what it actually delivered.
How the training works:
Every exercise hides a code that only appears once the task has really been solved. You enter that code and see straight away whether your solution is right. If you get stuck, a hint helps you on your way.
Every step ends with a short quiz in the course itself. It works without any technical setup, so you can test yourself again later.
Whether the code is good enough is not up to the agent, nor to the participant's gut feeling. A set of automatic checks decides, on test coverage, maintainability and security. Hard numbers instead of a hunch.
What you bring
Ready to make your developers work more efficiently?
Tell us briefly how big your team is and where it gets stuck today. We propose a day built around you, at your office, and we also say when a kata is not what you need.