Glossary

AI governance

AI governance is the set of rules, roles and checks an organisation uses to decide how AI may be used. For AI agents, it means deciding what each agent may do and who approves which steps, and keeping the evidence.

In a run

The rules become things that happen in a run. An agent gets a credential for one job. A step outside it is refused. A sensitive step waits for a named person, and every decision goes on the record.

Our word for it

Limits you can see, and a record you can check.

Our position

Rules on paper do not stop an agent. Limits that apply on every call do. That is why we treat this as one part of a complete platform, not a product of its own.

Ares gives each agent credentials and an access policy, a controlled path to other agents and tools, a place in a workflow with human approval, inspection of what passes, and a record. Each part supports one of five guarantees, and each guarantee states its limit. We also build the solutions that run on it, with you.

What you still decide

Ares does not write your business rules. You approve them, and you decide which steps are sensitive. The record shows what agents did, not whether a rule was wise.

Ares also does not certify your organisation. For each piece of public guidance we map, every row says what Ares does today, what needs configuration and what it does not do.

See the guidance mappings, or run the sandbox to see a rule refuse a step.

Related terms

  • GuardrailsAI agent guardrails are the checks that stop an agent from taking an action, or passing on content, that it should not.
  • Control planeAn AI agent control plane sets and enforces the rules for agents, apart from the agents that do the work.

Tell us about one workflow.

We will say whether an agent should do it, and what the first step costs.

A senior member of the team replies within two business days.