Identity and delegation for AI agents

Metorial ties every agent action back to the agent and the person who delegated its access. You can always see who did what on whose behalf, with every grant logged and reversible.

Get startedBook a demo
platform.metorial.com
A screenshot showcasing the Identity + Delegation feature of Metorial.

Identity + Delegation

Know which agent acted, and on whose behalf

When agents act on people's behalf, the hard question is accountability: who actually did this, and who authorized it? Metorial ties every agent to a real identity and records the delegation behind its access. Each action traces back through the agent to the user who granted it, and every step is auditable, bounded, and reversible.

  1. 01

    Trace the full chain.

    Every action ties back to the agent that took it and the person who delegated the access. You can always see who did what on whose behalf.

  2. 02

    Audit the authorization.

    When a user grants an agent access, the delegation is logged too. You can review not just what an agent did but who allowed it and when.

  3. 03

    Keep it bounded and reversible.

    An agent can never exceed the person behind it, and any grant can be revoked instantly. The trail is backed by real control.

How the trail works

From an action to the person who authorized it

The accountability trail in Metorial connects every action to an agent and the user behind it.

Attribution

Action to agent

Delegation builds on agent identification. Every action is attributed to a specific agent rather than anonymous traffic.
Authorization

Agent to user

Each grant ties the agent back to the user who delegated it, and delegation can require approval through policies and access control.
Record

Logged and reversible

Both actions and the grants behind them are recorded in audit logs, and any grant can be revoked at any time.

What's on the record

A complete accountability trail

  1. Actions attributed to an agentEvery
  2. Grants loggedEvery
  3. Access beyond the userNever
  4. Revoke accessInstantly

See which agent acted. See who authorized it. Audit every grant. Revoke it the moment you need to.

FAQ

Answers to common questions about identity + delegation and how they fit into governed AI agent infrastructure.

  1. Can I tell which agent did something, and on whose behalf?
    Yes. Every action traces back through the agent that took it to the user who delegated the access. You can always see who did what on whose behalf.
  2. Yes. Every time a user grants an agent access, that decision is recorded in audit logs. You can review not just what an agent did but who authorized it and when.
  3. Delegation is how a user grants an agent access to a tool or integration. Agents start with no access, and every permission is something a user explicitly delegates, which is what makes the authorization trail possible.
  4. No. An agent is bound by its user's permissions, and a sub-agent can never do more than its parent agent. Delegation can't be used to escalate access.
  5. Yes. Delegation can be configured to require approval before access is granted. Sensitive access goes through a checkpoint that's also recorded.
  6. Yes. Because each agent has its own identity, access can be revoked instantly on a per-agent basis if an agent is misbehaving or compromised.

Always know who did what, on whose behalf

See how Metorial traces every agent action back to the agent and the user who authorized it, with full auditing and control. Book a demo, or get started.

Book a demoGet started