AI Identity Model

How AuthSpoke treats every AI agent as a first-class enterprise identity.

In AuthSpoke, an AI agent is an identity — modeled with the same seriousness as a human or service identity. Nothing autonomous is anonymous.

Every agent has#

A unique identity and id
An owner, business unit and department
A lifecycle state (DISCOVERED → REGISTERED → APPROVED → DEPRECATED → RETIRED)
An authentication method
A continuously computed trust score
Relationships to models, MCP servers, systems and data
A complete audit trail

Lifecycle#

Identity is created at registration (or promoted from discovery), governed through approval, and eventually deprecated and retired. Each transition is an authorization decision and an audit event, so an agent's history is always reconstructable.

Roadmap#

The identity model is designed to grow into issued credentials: client id, OIDC identity, JWT, service identity and optional certificates per agent — plus agent-to-agent delegation and cross-organization federation. The directory you build today is the foundation for that.

See AI Agents for the working model and API.