Microsoft launches Copilot Cowork with Anthropic’s Claude, adds AI tasks across M365,

Microsoft’s rollout of Copilot Cowork marks the first time Anthropic’s Claude Cowork engine has been embedded directly into the 365 Copilot stack, enabling the AI to orchestrate “long‑running, multi‑step tasks” across Outlook, Teams, Excel and other core apps, according to the company’s own announcement and coverage in Fortune and CRN [Fortune; CRN]. The integration is positioned as a cloud‑powered “agentic automation tool” that takes a natural‑language request, drafts a plan, and then executes the plan by pulling data from emails, calendar events, shared files and spreadsheets without further user intervention [VentureBeat; The‑Decoder].

The new capability is part of Microsoft’s broader “E7” AI suite, a collection of enterprise‑focused tools that the firm says will boost workplace productivity by surfacing insights and automating routine work [CRN; mezha.net]. Within the E7 portfolio, Copilot Cowork is the only component that can act autonomously across multiple M365 services; other E7 modules remain confined to single‑app assistance. Microsoft’s product lead described the feature as a “cloud‑powered AI agent” that runs in the background, continuously monitoring context and updating results as new information arrives [VentureBeat].

From a technical standpoint, Claude Cowork supplies the planning and execution engine, while Microsoft provides the connective tissue that maps the agent’s actions to the APIs of Outlook, Teams, Excel, Word and PowerPoint. The partnership with Anthropic was highlighted as a “close collaboration” to ensure the agent can handle “multi‑step, long‑running tasks” such as generating reports, reconciling data across spreadsheets, and drafting meeting summaries [The Verge; CNET]. Microsoft stressed that the agent operates under a governance layer that can be toggled on or off, a response to concerns that “un‑governed AI agents could become corporate ‘double agents,’” a risk the company says it mitigates for $99 per month per user [VentureBeat].

Enterprise customers will initially see Copilot Cowork in a limited preview, with Microsoft promising broader availability later in the year. Early adopters in the telecom and data‑mesh sectors are already testing the platform through a joint effort with Tech Mahindra, which is building an ontology‑driven AI framework on top of the same Microsoft‑Anthropic stack to accelerate digital transformation in the region [Tech Mahindra report]. The collaboration suggests that Microsoft intends to extend the agentic model beyond productivity tools into industry‑specific workflows, leveraging the same underlying Claude Cowork technology.

Analysts note that the move deepens Microsoft’s AI moat against rivals such as Google’s Gemini‑powered Workspace tools and Anthropic’s own enterprise offerings. By embedding Claude Cowork within its flagship productivity suite, Microsoft not only differentiates 365 Copilot from competing assistants but also creates a data‑rich feedback loop that can improve the model over time. The company’s strategy, as outlined in its E7 launch, is to make AI an “always‑on” collaborator that reduces friction in routine tasks while preserving user control through governance settings [CRN; mezha.net].

If the preview lives up to its promises, Copilot Cowork could redefine how enterprises allocate human labor, shifting many repetitive activities from manual execution to autonomous AI agents. However, the success of the feature will hinge on the robustness of its cross‑app orchestration, the effectiveness of its governance controls, and the willingness of corporate IT departments to adopt a $99‑per‑month licensing model for AI agents that operate with limited human oversight.