AI-Native Team Day
Your team has AI licences and nobody agrees on how to use them. One engineer ships three times faster, another produces code review cannot trust, and no one has written down the difference. In one day I give your team the working discipline: agent workflows, MCP, context engineering, correction protocols, and the review rules that keep AI output from quietly becoming tech debt. It also helps meet the EU AI Act's requirement that staff working with AI understand what they are using. They leave with working setups and a written playbook, not slides.
One day. Your team leaves AI-productive. Most companies bought AI tools and skipped the discipline. The result is uneven: real speed in some hands, review-blocking slop in others, and no shared standard for which is which. The teams getting compounding value are not the ones with better licences, they are the ones with rules about context, correction, and what an engineer must verify before merging. I have spent two years building exactly that discipline, on five production systems and my own agent infrastructure, and this day transfers it.
Every deliverable, upfront.
Working session on your real repository, not a toy example
Agent workflows: when to delegate, when to pair, when to refuse
MCP servers and tooling, so agents reach your systems safely
Context engineering: what to load, what to leave out, why it decides quality
Correction protocols: naming a failure mode instead of re-prompting hopefully
Review discipline: what a human must verify before AI-written code merges
A written playbook your team keeps, tailored to your stack
Follow-up call two weeks later to fix what did not stick
How it runs.
Before, Prep: a short call plus repository access so the day uses your code, your stack, and your actual friction rather than a generic demo.
Morning, Foundations: how the leverage actually works, hands on keyboards. Agent workflows, context engineering, and the failure modes everyone hits in the first month.
Afternoon, Your codebase: we take real tickets from your backlog and work them with the discipline applied, then agree the review rules your team will actually enforce.
After, Playbook and follow-up: written rules delivered within two days, plus a follow-up call two weeks later once reality has tested them.
Often, What's next: teams frequently bring me on afterward to implement what we scoped together, on a day rate or a fixed engagement. The workshop is a low-risk way to find out if we work well before anything bigger.
This is for you if…
Engineering teams with AI licences and no shared standard
Tech leads seeing quality drift since AI tools arrived
Companies that want the speed without the review backlog
Ready to start?
On request one day, up to 10 engineers · 1 day · No pitch deck.
German SMEs can often fund part of this
As team training, this can be eligible for Bildungsscheck NRW (50% of the fee, up to €500 per person) and other Weiterbildung programs. You apply, eligibility depends on your company; I'll point you to the right programme on the call.
01Does this help with the EU AI Act?
Yes. The EU AI Act requires that people who work with AI systems have adequate AI competence, and this day is a concrete, documented step toward that: your team leaves understanding the tools, the risks, and how to catch a wrong answer, with a written playbook as the paper trail. It is not legal advice, but it is exactly the kind of practical enablement the competence requirement expects.
02How is this different from just buying the team Copilot licences?
Licences give access, not discipline. The teams getting compounding value have shared rules about what to delegate, how much context to load, how to correct a wrong answer, and what a human must verify before merge. That is what this day transfers, and it is the part no tool ships with.
03Is it slides or hands-on?
Hands-on, in your repository. We use your stack and your backlog, because generic demos never survive contact with a real codebase. Slides would be faster to prepare and worth far less to you.
04Remote or on-site?
Both work. Remote runs fine over video with shared sessions. On-site in NRW is straightforward and often better for a first day, since the informal conversation between blocks is where the real objections come out.
05How many people can attend?
Up to ten engineers at the standard price. Beyond that the hands-on part stops working and we should split it into two days rather than pretending a bigger room is the same thing.
06What does my team need to prepare?
Repository access before the day, working AI tooling on their machines, and a few real tickets from the backlog. That is deliberately light: the point is to work on the problems you already have, not to prepare for a workshop.
Still unsure whether this is the right shape of work?
Book a call