What is non-human identity management for AI agents?

Non-human identity (NHI) management for AI agents means treating every autonomous agent as a distinct identity with its own credentials, permissions, audit trail, and lifecycle — rather than having agents share human user accounts or service account credentials.

Most AI agents today run under shared identities:

• Shared service accounts: 10 agents share one database credential — when one misbehaves, you can't tell which
• Developer credentials: Agents run with the developer's personal API keys — far more permissions than needed
• No identity lifecycle: Agents are created and destroyed dynamically, but their credentials persist — orphaned credentials accumulate
• No MFA: Humans use multi-factor auth, but agents authenticate with a single API key that can be stolen

Exogram assigns each agent a unique namespace identity evaluated at Gate 0. Every action is attributed to a specific agent identity. The Trust Ledger records which agent did what, when, and with which credentials. When an agent is decommissioned, its namespace is frozen — all pending actions are rejected. Identity management isn't optional for AI agents. It's the foundation of governance.