AI Internal Control: AI System Inventory and Agent Register

AI System Inventory and Agent Register is a governance tool from the AI Assurance Institute for recording and overseeing an organisation’s AI systems and AI agents in a single, structured register. It keeps one record per AI system. Each record is linked to the system’s stakeholders, inventory users, use cases, misuse cases, capabilities and agents. The aim is to give organisations a consolidated view of their AI portfolio as a foundation for accountability, risk classification and wider AI governance.

Each system record captures core attributes: name, purpose, provider, version, accountability, watermarking, the use of personal data, and a first-pass EU AI Act risk classification. An agent registry adds agent-specific detail. This includes each agent’s autonomy level, its Model Context Protocol (MCP) allowlist of permitted tools, and any high-risk combinations of tools.

Records move through a draft, review and approve workflow, so entries are checked before they are accepted into the inventory. Operators manage the register through a dashboard and a governance map that show how systems, agents and use cases relate to each other. The register can optionally sync with the institute’s AI Analyzer. It also supports federated export and import, so inventories can be shared or consolidated across teams or entities. It is one of several solutions the institute offers alongside impact assessment, ISO/IEC 42001 and AI quality management platforms.