Microsoft announced Windows 365 for Agents, a Cloud PC platform that runs autonomous AI agents inside enterprise-managed Windows environments, according to a Windows blog post dated January 22, 2026. The platform lets organizations direct agents with natural language to interact with applications, browsers, files, and enterprise systems, ITSecurityNews reports. Per reporting, builders can automate workflows that rely on applications and systems without APIs, including legacy and UI-based environments. The Windows blog describes a hosted-on-behalf-of (HOBO) architecture that uses single-instance Azure virtual machines managed through Microsoft Intune, secured with Microsoft Entra ID, and connected via reverse connect transport. ITSecurityNews reports the service is available in public preview.
What happened
Per the Windows blog dated January 22, 2026, Microsoft introduced Windows 365 for Agents, an extension of Windows 365 that enables organizations to run autonomous AI agents on Cloud PCs under enterprise controls. According to ITSecurityNews, the platform allows agents to be directed with natural language to interact with applications, browsers, files, and enterprise systems. ITSecurityNews additionally reports the platform is available in public preview and can automate workflows that rely on applications and systems without APIs, including legacy and UI-based environments.
Technical details
Per the Windows blog, Windows 365 for Agents is built on the service's existing cloud-PC foundation and a "hosted on behalf of" (HOBO) architecture. The blog describes these core components:
- • Microsoft Entra ID for authentication and Conditional Access - • Microsoft Intune for device management and policy enforcement - •single-instance Azure virtual machines running in Microsoft subscriptions
- •reverse connect transport to link Cloud PCs to Microsoft management
These elements are presented as the mechanisms Microsoft uses to provide identity, policy, security, and connectivity for agent workloads on Cloud PCs.
Industry context
Editorial analysis: Organizations that deploy autonomous agents into enterprise systems typically face requirements for centralized policy, auditing, credential handling, and network isolation. Vendors framing agent execution as a managed Cloud PC workload follow an industry pattern that shifts enforcement and telemetry into a centrally controlled runtime, which can simplify governance for UI-driven or legacy-system automation compared with ad hoc on-device agents.
What to watch
For practitioners: monitor these indicators as the preview evolves: integration with third-party LLM and agent frameworks, granularity of permission models for UI and data access, audit and telemetry features for agent actions, secret and credential handling across Cloud PCs, and performance/cost metrics for scale. Also track how vendors and enterprises reconcile policy-driven agent execution with existing RPA and automation tooling.
Scoring Rationale #
A major vendor offering a managed runtime for AI agents matters to enterprise practitioners evaluating secure, auditable agent deployments. The story is product-level rather than frontier model research, so it is notable but not industry-shaping.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.