Governed AI for teams that have to prove it.
Learn Agentic AI helps regulated organizations adopt AI without losing control of it. The focus is narrow on purpose: banks, research institutes, and high-security environments, where an AI action needs an owner, a log, and a boundary it cannot cross.
The work is led by Istvan Szukacs, who builds data platforms, AI systems, and cloud infrastructure. He built infrastructure at Amazon, delivered the data warehouse behind League of Legends for 100M+ users, and migrated a 5PB data warehouse to the cloud at Amadeus. He teaches Databases and Data Management at the Frankfurt School of Finance and Management and writes about production systems at vectorian.be.
The approach
Governance comes before technology selection. We define the control plane, the six requirements, and the autonomy tiers first. Then the tooling and the pilots fit inside them rather than the other way around.
The public material stays concrete but general. The detailed control architecture, the gateway configuration, and the evidence model for your regulator are built with your teams during an assessment.
The standard we hold
Autonomy is earned with evidence, not granted by default. A use case moves up the ladder only after its controls are proven in production. That is the difference between an AI pilot you can defend and one you cannot.