AI governance platform
The system that lists AI agents, limits them, routes risky actions to a person and records everything.
Platform overviewGlossary
What do the main terms in AI governance mean?
Short definitions in plain words. Search for a term, or browse them all.
- Searchable
- One sentence each
Terms
AI agent
Software that uses an AI model to take actions, not only to answer questions.
Agent register
One list of every AI agent, with what it does and who owns it.
Know: the registerShadow AI agent
An agent running in your systems that the responsible people do not know about.
Policy
The rules for an agent: what it may reach, do alone and spend.
Authorize: policyApproval tier
One of three levels for an action: acts alone, needs approval, or never allowed.
Named approver
A specific person with the right role who decides a high-risk action before it runs.
Operate: approvalsHuman in the loop
A design where a person reviews or decides certain AI actions before they take effect.
Audit trail
A record of every AI action: agent, action, approver, time and cost. In MFDIO it is append-only.
Prove: the audit trailWork Unit
MFDIO's measure of AI work. One budget in Work Units replaces scattered usage prices.
Work UnitsRecord
The name MFDIO uses in the product for its audit trail.
The audit trailEmergency stop
A control that pauses every active agent and blocks AI actions at once.
MCP
Model Context Protocol, an open standard for connecting AI assistants to tools and data.
No term matches.
The four steps in terms