Microsoft said Nov. 17, 2026, that Windows 365 for Agents is now eligible for Copilot Credits, bringing the company’s common AI billing currency to Cloud PC agent workloads.

Rapid News Brief: Release stages

Microsoft’s announcement says on-premises access is in preview. Persistent Cloud PCs are planned for private preview, while Microsoft Foundry integration is slated for preview early next year. A Microsoft Copilot harness preview in Copilot Studio is due in December. The underlying service became generally available earlier this year.

A pilot, not a companywide rollout

In a separate Nov. 17 internal deployment report, Microsoft described a live pilot in one Business Operations unit, with wider adoption still planned. The account says process experts first record how work should be performed; the BPO AI Toolkit turns that knowledge into instructions, policies, reusable agent capabilities and an execution plan.

The company also reports administrative friction: provisioning additional Cloud PCs required manual tenant-administration steps. Teams are working with Microsoft 365 FinOps on usage limits and cost allocation. Its example of restricted permissions allows an order-entry agent to create an order, but not cancel or delete one. People review exceptions with evidence of the steps already taken. These are descriptions of Microsoft’s implementation, not measured results for outside customers.

How the pooled model works

Microsoft’s Jan. 22 architecture article describes a pooled model in which agents check out a Cloud PC for a task and return it for reuse, rather than receiving permanent one-to-one assignments. It says the service uses Azure virtual machines with Intune management and Entra identity systems. Administrators can apply provisioning policies, applications, settings and security baselines through the same Intune console used for physical devices. The design also allows a person to take control during execution, handle a decision or provide credentials, then return control to the agent.

These design descriptions explain the service’s allocation and handoff mechanics; they are not an independent test of reliability, security or cost savings.