Agent identity is the treatment of an AI agent as a first-class non-human identity with its own credentials, entitlements and lifecycle — so that its actions can be authorised, constrained and attributed rather than hidden behind a shared service account.
What is agent identity, and why does it matter?
Agents deployed with a shared service credential inherit whatever that account can reach, which is typically far more than any individual user, and their actions become indistinguishable from one another in logs. Giving each agent a federated identity allows entitlements to be scoped, delegation to be explicit, actions to be attributed, and access to be revoked when the agent is retired. It also makes the essential constraint expressible: an agent acting on behalf of a user must never exceed that user's own entitlements.