AI Systems
Each AI system you register here gets a model card, AI BOM, risk register link, and continuous oversight log — the artifacts a notified body asks for.
Risk-class wizard — 3-minute system registration
1
Does the system make decisions that significantly affect natural persons?
Examples: credit scoring, hiring, healthcare triage, education evaluation, biometric ID, border control, critical infrastructure. → likely high-risk under Annex III.
2
Is the output user-facing chat or content generation?
Generative systems require transparency obligations (Article 50) — synthetic-content labelling, deepfake disclosure. Independent of risk class.
3
Is it a general-purpose AI model (foundation model) at deployment?
GPAI obligations (Article 51–55) — model card, training data summary, copyright policy, computational resource disclosure if >10²⁵ FLOPs.
4
Result: pick a risk class
High-risk (Annex III)
Limited (Article 50)
Minimal
GPAI
REGISTERED SYSTEMS
Open risk register
No AI systems registered yet
Use the Agentics SDK or the risk-register API to register a system. Each system gets a model card surface, an AI BOM (CycloneDX), continuous oversight, and a one-click Article 26 deployer pack.
SDK reference Add a first risk