The governance policies restricting which foundational LLMs, fine-tuned models, and inference endpoints an agent is permitted to call.
Restricts unapproved or high-cost model usage, ensuring high-risk tasks use approved models with required privacy guarantees.
A medical triage agent is enforced to call only HIPAA-compliant, private-endpoint models rather than public commercial endpoints.
Prevents data leakage to public models and controls model inference expenditure.
EnterpriseACP enforces model pinning and endpoint validation in the runtime control plane.