Microsoft is giving agents their own Cloud PC capabilities to help you with crucial tasks
Microsoft wants companies to be able to give their AI agents their own identities, and that includes allocating them a Cloud PC.
Microsoft is rolling out new capabilities to empower AI agents with their own Cloud PC resources, enabling them to perform crucial tasks as if they were human workers. These AI agents will be managed through Microsoft Intune and Entra, receiving unique identities and permissions. The objective is to improve general Cloud PC performance and reliability while providing AI agents with deeper system-level access to legacy and unsupported applications, rather than requiring companies to rebuild or migrate their applications entirely.
Through this initiative, AI agents will be able to interact with real Windows applications, conduct QA testing, and generate documentation within a secure, enterprise-governed Windows 365 for Agents environment. Microsoft is also introducing Direct Stream technology to enhance responsiveness, reliability, and performance, allowing users to connect to their Cloud PCs within seconds, making them feel much more like conventional, local machines.
Additionally, early customers report that AI agents can utilize their own Cloud PCs to perform QA testing and generate documentation. Windows 365 for Agents will be eligible for Copilot Credits, a unified currency applicable across Microsoft's AI offerings to facilitate easier billing and spend visibility. The company emphasizes that Cloud PCs are managed similarly to those for human users, with their own identities that can be configured to be more restrictive than human identities.
Bhavya Chopra, Product Leader for Windows 365, highlighted the company's vision for the future of computing, stating that it requires a familiar, flexible workspace that can accommodate both people and their AI agents, regardless of where work takes place.
Written by urgent.news from TechRadar's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.