Use cases for asynchronous authorization
AI agents often need to perform actions that are long-running, are executed at a later time, or involve sensitive operations. For example:- Transactional approvals: An agent arranging a business trip might find a flight and hotel, but needs the user’s final approval to complete the purchase.
- Accessing sensitive data: An agent might need to access a protected user’s health records to answer a query.
- Executing high-privilege tasks: A DevOps agent might need permission to deploy a new version of an application to a production environment.