Resources · Glossary
AI agent policy, in plain words.
The terms used across Gaten, explained without jargon.
- AI agent
- An AI system that can take actions through tools — send an email, issue a refund, change a record — not only answer questions.
- Allow
- A policy outcome: the action is safe to run automatically, and it goes ahead.
- Approval
- A recorded yes or no from a named person on a single held action, before it runs.
- Audit trail
- A record that can't be edited after the fact of what each agent tried to do, what was decided, by whom, and why.
- Deny
- A policy outcome: the action is never allowed, and the agent receives a clear reason.
- Dry-run
- Testing a new rule against past agent activity to see what it would have done, before switching it on.
- Escalation
- What happens when an approver doesn't answer in time: the action moves to someone else, or is safely denied.
- Human in the loop
- A person must approve an action before it happens.
- Policy
- A set of rules that decides whether an agent action is allowed, denied, or needs approval.
- Policy gateway
- The checkpoint that sits between an agent and its tools, so every call is checked before it reaches a real system.
- Policy version
- Each saved change to a policy, so you always know which rules made a decision.
- Tool call
- The moment an agent asks to use a tool — the point where the rules are checked.


Every agent action, checked first.
From your first support bot to a fleet of autonomous agents, talk to an expert about putting your agents under policy with Gaten.