An AI agent needs to retrieve customer information from an API. What should happen between the agent and the API?

An AI agent should not directly access an enterprise API without appropriate controls between the agent and the backend system. The request should pass through a controlled tool or application layer that validates the agent's requested operation and parameters. Authentication and authorization should verify the requesting identity and ensure the agent has access only to the required customer data. The API should validate inputs, enforce business rules, and return only the information permitted for that request. Logging, monitoring, rate limiting, and error handling should also be considered so that API activity remains traceable and controlled. How would you design the flow from Agent → Tool → API → Enterprise System?