Someone who owns the outcome.

The problem this solves

Most AI budgets are not lost on the wrong build. They are lost because nobody owns the programme.

The pattern is consistent: a promising pilot stalls between two departments. The compliance question surfaces after the build instead of before it. A vendor delivers something technically real that never gets integrated. Everyone agrees AI matters and nobody is accountable for shipping it. A year passes; the demo still demos.

That is a leadership gap, not an engineering one — and hiring your way out of it is slow and expensive. A head of AI in Germany is a six-figure salary plus employer costs, the search takes months, and at the point where you most need direction you can least justify the headcount.

What the role actually does

• Decides what is worth building. Which workflow pays for itself, which one looks impressive and returns nothing, and which should not be automated at all.
• Owns delivery. Scope, sequence, dependencies and the hard call when a pilot should be killed rather than extended. One accountable person, not a steering committee.
• Runs the compliance lifecycle in parallel. AI-Act classification and technical documentation, and the GDPR work that rides along with it, produced while the system is built — never reconstructed afterwards, which is where it becomes expensive.
• Holds vendors and internal teams to a standard. Someone in the room who has shipped this before and can tell a real architecture from a convincing one.
• Leaves documentation behind. Every decision recorded so the programme survives the engagement — and so your team can take it over rather than depend on me.

Why me

Twenty years in enterprise IT, including BMW and Oracle. Delivery at a scale where a missed dependency costs real money, and where “it works on my machine” was never an acceptable answer.

Certified Project Management Professional (PMP), issued by the Project Management Institute. The AI part of this role is the part everyone talks about; the delivery part is the part that decides whether anything ships.

Four production SaaS products, designed, built and operated solo — Argumentree, ArgumenTroupe, AI-Agentree and Argumentree.AI. Multi-tenant, real-time, live. Not a portfolio of mockups.

One of those products is an EU-AI-Act governance platform. The compliance lifecycle is not something read about in a guide here — it is a system built, shipped and operated: decision traces, agent identity, audit lifecycle.

The combination is the point. Enterprise delivery discipline is common. Hands-on AI engineering is common. Both, in one accountable person who can also write the code, is not.

What this is not

• Not a consultant with a deck. The deliverable is a system in production and the evidence that it is compliant — not a recommendation.
• Not a project manager by the day. Fixed monthly scope, agreed up front. Nobody bills you for attending meetings.
• Not a member of your staff. An external supplier under a framework agreement — advising, reviewing and holding the standard, not line-managing your people. Own premises, own tools, no badge and no seat on your org chart. It keeps the engagement clean for your side as much as mine (see how we contract).
• Not a lock-in. The explicit goal is that your team can run the programme without me. Documentation is a deliverable, not a favour.
• Not legal advice. AI-Act and GDPR work here is engineering: classification, technical documentation, and the technical input your DPIA and data-protection officer ask for.

How it works

A fixed monthly commitment, scoped before it starts. No day rates, no timesheets — you are buying an owner for the outcome, and billing by the hour would put the incentives on the wrong side of the table.

Concretely, the engagement is defined by what it produces each month, not by availability: a maintained AI roadmap, a programme review with a written decision record, the AI-Act classification kept current for the systems in scope, and vendor or architecture reviews as they come up. It runs for an agreed term — typically six or twelve months — and renewal is a deliberate decision, not an automatic rollover.

Engagements begin with the free AI-Act check and a conversation about where AI actually pays in your business. If a build follows, it runs as its own fixed-scope tier — assessment, pilot or production — so leadership and delivery are never bundled into one unaccountable number.

The studio is capacity-bound by design, so this tier is limited. That is also what makes it worth having.

What it costs

From €3,900 a month. For comparison: a permanent head of AI is a six-figure salary plus employer costs and a months-long search, and a mid-level AI engineer alone runs €70,000–90,000 a year. The full ladder, from the free check to a production deployment, is on the services page.

Questions

How is this different from an AI consultant?

A consultant delivers a recommendation and leaves. A delivery lead owns the outcome and is still there when the thing has to ship, integrate and survive an audit.

Do you work with our existing developers?

Usually, yes. The role brings direction, delivery discipline and the compliance lifecycle to the team you already have — it does not replace it.

Can you also build what you plan?

Yes, and that is why the plans are buildable. Building runs as a separate fixed-scope tier, and this engagement stands on its own if you would rather build in-house.

How is this contracted — is it not just employment by another name?

No, and it is structured so that stays clear for both sides. It is a Werkvertrag under a framework agreement, defined by monthly deliverables rather than days or availability, for a fixed term with an explicit renewal. I work from my own premises with my own equipment, carry my own liability, run several clients at once, and hold no managerial authority over your staff. Full detail under how we contract. Your counsel takes the legal view; the structure is built to make that view an easy one.

How quickly can this start?

Far faster than a hire — there is no search, no notice period and no onboarding ramp. Subject to capacity; the studio deliberately runs few engagements at a time.

What happens to the work if the engagement ends?

It stays with you. Documentation, decision records, source and infrastructure are yours throughout — written so your team or your next supplier can pick it up. See data sovereignty for where the systems themselves run.

We are not sure the AI Act applies to us.

Start with the free check. It returns an applicability answer, a risk classification and the duties that follow. From 2 August 2026 most obligations apply.