Formerly known as Azure AI Services or Azure Cognitive Services is a unified collection of prebuilt AI capabilities within the Microsoft Foundry platform
Hello EV Admin,
Welcome to Microsoft Q&A and Thank you for reaching out.
The observed behaviour is consistent with a scenario where the Agent Application completes provisioning at the control‑plane level while the associated agent identities continue to finalize in the background. In such cases, the top‑level provisioning state may show as successful even though the dependent identity components have not yet fully transitioned.
Agent identity provisioning is an asynchronous process. Agent identities and their related blueprints are automatically created and managed through Entra ID, with a lifecycle that is independent of the parent Agent Application resource. Until this identity lifecycle completes, dependent scenarios—such as Agent‑to‑Agent orchestration—may encounter temporary authorization issues, including 401 responses. Management operations on the Agent Application can also remain restricted during this transition period, even when the service principals already exist and are enabled.
Agent identities are fully platform‑managed and cannot be manually forced into a new provisioning state through update or patch operations. If identity provisioning remains incomplete for an extended period, the following practical steps may help move forward:
Please consider to see if the following troubleshooting steps helps:
- Please allow additional time for background identity reconciliation to complete.
- Validate that the required RBAC assignments are present at both the Agent Application scope and the project scope.
- Confirm that the Agent Application identities exist and are enabled in Entra ID.
- Then recreate the affected Agent Application to allow identity provisioning to restart cleanly.
- Kindly avoid repeated update or delete attempts while the identity state is still transitioning.
If the behaviour continues after these steps, deploying the same configuration in another supported region can help determine whether the behavior is region‑specific. Azure AI Foundry Agent Service availability, supported features, and preview capabilities can vary by region, and infrastructure readiness may differ.
When using Agent Service features, all dependent resources must reside in the same region. Creating a new Foundry resource and Agent Applications in an alternate supported region may allow identity provisioning to complete successfully and unblock Agent‑to‑Agent orchestration.
References:
Manage agent identities with Microsoft Entra ID - Microsoft Foundry | Microsoft Learn
Quotas and limits for Microsoft Foundry Agent Service - Microsoft Foundry | Microsoft Learn
Thank you!