Docs

Appearance

Platform Overview ​

This section tells how Modulos works: the objects, the procedures, the permissions, and the integrations for AI governance. Each subsection below is about one module of the platform.

Organizations

Workspace boundary, members, settings, usage, and shared libraries

AI Registry

AI Systems, the unit of governance scope for one deployed system or use case

Governance

Frameworks, Requirements, Controls, Evidence, and Reviews

Policy Center

Organization policies, their versions, approvals, and acknowledgements

Risk

Organization taxonomy, AI System risks, quantification, and risk limits

Runtime Inspection

Sources, Metric Tests, Agent Tests, schedules, results, and remediation loops

Integrations

API tokens, Sources, Scout Connectors, and access to external data

AI Agents

Scout, the Evidence Agent, the Control Assessment Agent, and the Risk Agent

Vendors

Third parties and their documents

How the platform is organized ​

Modulos uses a small set of objects. The relations between them make each governance step traceable from Requirement to Evidence.

  • Organizations are the top-level workspace. They set the members, the roles, the defaults for the full organization (language, currency, risk taxonomy), and the functions that are available. Each user belongs to one organization only.
  • AI Systems are the unit of governance scope, and they are in the AI Registry. An AI System is usually one deployed AI system or use case. It contains frameworks, Controls, Evidence, risks, and test results.
  • Governance is the compliance audit trail. Frameworks contain Requirements, Controls satisfy them, and Evidence shows that the Controls are done. Logs record each status change, but a framework update that archives a Requirement makes no log entry. One Control can satisfy Requirements of the EU AI Act, ISO 42001, and the NIST AI RMF at the same time.
  • Policy Center holds the policies of the organization, their versions, and their approvals. You can link published Policies to Controls.
  • Risk is quantitative risk management. Organizations set a risk taxonomy with categories and impact scales. AI System teams then identify each risk, give it a monetary value, and monitor the exposure against risk limits.
  • Runtime Inspection is continuous validation. Connect data Sources, write tests for properties that are important for governance (bias, performance, cybersecurity), and schedule them. Then map the tests to Controls. A failed test sets a mapped Executed Control back to Not Executed.
  • Integrations give authenticated access to external systems. Sources that you attach to an AI System connect external systems that Metric Tests, Agent Tests, Scout, and the Risk Agent read. The table in Supported Source types shows which of them read each type. User-level Scout Connectors (Google Drive, GitHub, Atlassian, Bitbucket) let Scout read connected sources during a chat.
  • AI agents help with the work: Scout answers questions and finds information, the Evidence Agent and the Control Assessment Agent help with Evidence and Controls, and the Risk Agent quantifies risks.
  • Vendors record third parties and their documents.

The left sidebar shows the primary areas at the top: AI Registry, Organization, Vendors, Policy Center, and Support. Below them, the sidebar shows the pages of the area that you are in, with the name of the area above them. In an AI System, the name of the AI System is above its pages. AI Registry above that name goes back to the AI Systems page. A small arrow button at the top of the right edge of the sidebar makes the sidebar narrow or wide.

Availability ​

The navigation can hide modules because of the setup and the subscription of the organization.

If a page in the docs is not in your workspace, speak to an Organization Admin about your access.