A single record that links what an agent can do to what could go wrong.

Create a workspace, describe your organisation and invite colleagues. Owners, admins, members and AI developers each get the access their role needs, and every change is written to an audit log.

Record each agent: what it is for, where it is deployed, which model and tools it uses, which systems it can reach and what it can do without human approval. Versions keep a history of how its capabilities change.

The risk engine will map an agent's permissions to concrete loss scenarios, score the exposure, explain its reasoning and recommend controls. What-if simulations show how a configuration change would move the result.

Upload existing policies to extract key terms, limits and exclusions, compare them with identified risks and see products that may be relevant to discuss with a licensed broker or insurer.

Configuration changes will be detected and reflected in the risk picture, with alerts when a change raises an agent's risk level.

When something goes wrong, record the incident against the agent and version involved, collect evidence and produce a report suitable for notifying a claim.