Governance, as the fast lane to production.
CognaOne is one platform for AI systems, agents, risks and controls, with evidence ready for regulators. You start where your AI programme is today and build on the same record.
Four stages, one record#
| Stage | What you do | Guides |
|---|---|---|
| Start | Connect platforms read-only, register what teams declare, import an existing register and give every AI system an owner | AI inventory, People |
| Prove | Rate risk against your risk appetite, map controls to CPS 230, CPS 234 and FAR, and collect evidence | Risk, Controls, Assurance |
| Scale | Bring agents into production with defined rights, a person on every high-risk decision and every decision on record | Agent authorisation, Security |
| Evolve | Map new obligations, reassess systems when they change and keep the evidence current | Discovery, Reporting |
Three lines, one platform#
Each line sees the same AI systems, owners, risks, controls and evidence. Nothing is copied between tools, so the first line's registration, the second line's challenge and the third line's test all refer to the same record. See Three lines, one record.
How it connects#
Platforms and data sources connect read-only through connectors and the CognaOne API. Agents connect through the API or MCP (Model Context Protocol) when you want CognaOne to decide before they act. See Connectors and discovery.
At the moment of action#
For agents in production, the same controls are used at the moment of action: CognaOne allows the action, routes it to a named person or blocks it. Try it in Agent authorisation.