The European AI Act is here. Secure your tech.
The AI Office assure your compliance with the EU AI Act. Don't let regulation scare off your customers.
We help you understand fast.
We tell you exactly what obligations your AI system falls under, in 10 questions.
We help your product team.
Your engineers should ship code, not navigate 144 pages of regulation. They ping us on Slack, we handle the rest.
We ensure your compliance.
We hand you a certified AI Act report to close enterprise deals and pass due diligence. When the regulation changes, we update your docs.
The EU AI Act ranks every AI system by risk. Your rank determines what you must do, and what you risk if you don't.
For the curious: read the original text
01 / The four risk levels
Four levels. One applies to your product.
Prohibited
In force since Feb 2025Social scoring, biometric surveillance, manipulation. Banned outright.
High Risk
Deadline: Dec 2, 2027Hiring, credit, healthcare, education, biometrics. Heavy documentation and oversight required.
Limited Risk
Deadline: Aug 2, 2026Chatbots, synthetic content. You must disclose that it's AI.
Minimal Risk
No deadlineMost AI tools, recommendation engines, spam filters. No specific obligations.
02 / Why it's complicated
Knowing your level is harder than it sounds.
Provider or Deployer?
Your obligations depend on whether you build the AI or just use it. Most startups don't know which one they are.
One product, multiple roles.
A system can be Low Risk today and High Risk the moment you add a feature that scores or categorizes people.
The text keeps changing.
The AI Omnibus (May 2026) moved a major deadline by 16 months. Staying current is a job in itself.
Not sure where your product stands?
You'll get your risk level, your obligations, and your deadlines, specific to your system.
03 / Pricing
We handle the work, and stay by your side as the regulation evolves.
04 / About us
