Understand
We find the one workflow where an agent pays for itself — and check where the AI Act and GDPR apply before a line of code is written.
AI agents · Berlin
Custom AI agents for your business — designed, built, and run by one engineer who ships. EU-AI-Act- and GDPR-compliant by construction, built in Berlin and hosted in the EU.
How we work
No slide decks that gather dust. We map the one workflow worth automating, ship a working agent, and keep it running — compliant, monitored, and yours.
We find the one workflow where an agent pays for itself — and check where the AI Act and GDPR apply before a line of code is written.
A working agent in weeks, not quarters. Built on a reusable chassis, so you pay for your problem — not our boilerplate.
Agents drift. We monitor, document, and hold the compliance line — governance built into the product, not bolted on after.
Why we're fast
Four production products share one engineering core — auth, tenancy, tracing, governance. Your agent starts on top of that core, not from an empty repository. That is how a boutique of one ships what agencies quote in quarters.
Proof of work
Not a portfolio of mockups — live, multi-tenant software in production. Including an EU-AI-Act governance platform and an AI research engine.
Structured argument-mapping and decision-making. Multi-tenant, real-time, in production.
Visit Argumentree →Panels of AI personas that debate a question from every angle — agreement, contrast, and edge cases surfaced.
Visit ArgumenTroupe →An AI-governance and EU-AI-Act platform: decision traces, agent identity, and the full audit lifecycle.
Visit AI-Agentree →An AI research engine: turns a question into a sourced, structured decision map, with arguments extracted, rated, and traced by AI.
Visit Argumentree.AI →EU AI Act
The Act has applied since 2 August 2026, and the Digital Omnibus moved only the high-risk duties — to December 2027. That is not more time; it is one fewer excuse. The classification is what tells you which AI ideas are worth building at all, and what each one costs to run properly. Every agent we build is classified, documented, and governed from day one — so the regulation becomes an advantage you own, not a wall you hit. And it never travels alone: almost every agent touches personal data, so GDPR rides with it — which is why we build for both at once.
What the AI Act requires, and how we build it in →
Data sovereignty & GDPR
The AI Act governs the system; GDPR governs the data — and almost every useful agent touches personal data. So we engineer data protection the same way we engineer compliance: from the first line, never bolted on. The model runs on an EU-resident endpoint, nothing is retained, and your data is never used to train it — and you choose how close to home the whole agent runs.
Bound by professional secrecy?
For lawyers, tax advisors and doctors, § 203 StGB doesn’t bend for AI: run privileged data through a consumer chatbot and you can breach it. Our agents keep that data on an EU-resident, zero-retention endpoint — or never let it leave your own hardware. The duty of confidentiality stays intact — the line between an AI you’re not allowed to use and one you are.
EU-hosted, EU-resident inference (Azure Sweden Central) — zero retention, never used for training. On request, in your own cloud or fully on-premise.
Data minimisation, access control, and technical and organisational measures built into the agent — not a policy written after the fact.
The right to erasure works because it is built: retention windows and archive-then-purge, the same data lifecycle that runs our own products.
A ready data-processing agreement, and the technical input your DPIA and your data-protection officer ask for — produced by the build, alongside the AI-Act evidence. Engineering, not legal advice.
Where it runs
From a managed EU endpoint to your own hardware — the same agent, four ways to deploy it. Pick what your data policy needs; move between tiers later without a rebuild.
Start here
A quick three-step check — your industry, your systems, and where the gaps are — with an instant compliance score and the duties that follow. Powered by AI-Agentree, our own EU-AI-Act platform.
Take the free check →Services
Fixed scope, not open-ended day rates. For context: a mid-level AI engineer runs €70–90k a year, and an agency build is quoted in six figures — and every tier below starts with the free check. Full scope and pricing →
Two minutes: does your project fall under the AI Act, and at what risk tier?
Two depths, one rung. Classify puts one named system’s AI-Act risk tier and duties in writing, flat €2,900, inside five working days. Clarity (from €5,900) covers the whole estate and ends in a build plan — and credits the Classify fee against it.
A working pilot on your own data and workflow, in weeks.
Hardened, monitored, and documented for audit.
A fractional lead who owns your AI initiatives and the compliance lifecycle — 20 years of enterprise IT, PMP-certified.
A team day: which use cases are worth building, and how the AI Act classifies them. Often the first step, not the last.
The scene
We’re also mapping who is building agents in Berlin — startups, MCP servers, and studios, competitors included. A machine-readable index, and a monthly note on what’s shipping. Launching soon.
Get on the list →Let’s talk
Tell me the workflow that’s eating your team’s time. If an agent fits, I’ll show you how — starting with the free AI-Act check.
or write to hello@agent.berlin
Who builds this
Twenty years in enterprise IT — including BMW and Oracle — and a certified Project Management Professional (PMP). Today: a solo engineer in Berlin who designed, built and operates the four products above, one of them an EU-AI-Act governance platform.
Hire agent.berlin and you work directly with the principal who owns the build end-to-end — no account managers, no offshore hand-off, every deliverable documented to outlast the engagement.
Twenty years of enterprise IT, and what happens if I’m unavailable →