Appearance
AI Agents Overview
Modulos has AI agents for governance work. The agents use the data of your AI Systems and your organization. The users of Modulos stay accountable for decisions and approvals.
Responsible AI
Modulos follows the principles of responsible AI when it develops and operates its platform.
- Modulos is a signatory of the European Commission’s AI Pact.
- Read our Code of Responsible AI.
What’s included today
Your organization configuration sets which agents are available. Modulos has these agents:
- Scout: a chat assistant. It answers questions about one AI System, a Draft, or all the AI Systems that you can read. It can read the Sources and Connectors that you select, and it can propose new AI Systems, tests, and files as Evidence.
- Evidence Agent: suggests the Controls for new Evidence, and a name and a description for one Evidence item.
- Control Assessment Agent: calculates the readiness of a Control from its Evidence and policies, and stores the result.
- Risk Agent: calculates a monetary value for a risk threat, with its assumptions and references.
- Agent Tests: checks that an AI agent does on the Sources of an AI System, against Pass Criteria in your own words.
How AI stays auditable
No agent output approves work, sets a Control to Executed, or sets a Requirement to Fulfilled. A failed Agent Test sets each mapped Executed Control back to Not Executed (Agent Tests). Modulos stores some outputs when a run completes, and you can run the agent again. Human in the Loop tells where a person makes the decisions.