AgentSpace Berlin
Sovereign, secure and controllable agentic AI
AI agents that act on their own, booking, filing, deciding, promise large gains for companies and public administration. They also raise new questions: who controls them, how are their actions traced, and how do they stay within the rules?
AgentSpace Berlin is planned as an innovation space for sovereign, secure and controllable agentic AI. As lead applicant, I coordinate eleven partners from industry and the public sector, including n8n, Cavorit, Technologiestiftung Berlin, Aiio and the Berlin Senate Chancellery. The proposal to the Federal Ministry of Research, Technology and Space (BMFTR), with approx. EUR 2 million requested, is under review.
The project applies the principles of my research on validity and uncertainty: explicit representations, traceable decisions and human control.