Many humans, many agents

Addressable team member

An addressable team member is an agent with a name, an owner, scoped permissions, a spend limit and an audit trail, administered from the same console that governs human users and assignable work the same way.

Control sits with both parties

When to use it

Any deployment where more than one person relies on the same agent.

What it prevents

Shadow agents that nobody owns, and the governance gap where an agent has broader access than any person on the team.

Rosettatranslation agent
OwnerMarketing, Elin
RemitDraft translations. Never publishes.
AccessContent library, read and write. Nothing else.
ReviewedEvery quarter, next one due
A name people can use in a sentence, and an owner to ask.

Who ships it

Sources

Related patterns: Permission parity · Agent registry · Attribution in the shared record

All agent design patterns

Review an agent feature with us30 minor send a message