Skip to main content
An AI Use Case is the approval boundary for a provider and model combination within a project. Rather than letting any agent call any model from any provider, you define use cases up front, and AI Agents are created against an approved use case.

An approved AI Use Case, showing its AI Provider, Permitted Models, and Approval Journey

​
Fields

  • Name (required)
  • Description (optional) — what the use case is for
  • AI Provider (required) — selected from your organization’s connected providers
  • Allowed Models (required) — the specific models permitted under this use case
  • Risk Rating (required) — the risk level assigned to this provider/model combination
  • Notes (optional)
  • Policy Set (optional) — a default Policy Set applied to agents created from this use case, unless overridden

​
Status

A use case can be draft, proposed, confirmed, active, deprecated, or rejected. Only confirmed or active use cases are available when creating a new AI Agent.

​
Restricted editing

Not every team member can edit every field. If you don’t have full access to a use case, you can still update its Notes and assigned Policy Set — provider, allowed models, and risk rating can only be changed by organization owners and admins. This keeps the approved provider/model boundary stable while still letting teams adjust guardrails as needed.