Microsoft has announced plans to integrate Anthropic’s AI agent technology into Microsoft 365, allowing users to delegate workplace tasks to autonomous AI systems that operate across documents, emails and collaboration tools.
The feature, called Copilot Cowork, will enable users to hand off tasks that the AI will convert into a step-by-step plan and execute across applications such as Microsoft Outlook, Microsoft Excel and Microsoft Teams.
Microsoft said the system will operate within the existing security and governance frameworks of Microsoft 365 while drawing on users’ work data to perform tasks.
Users will also be able to interrupt the AI mid-task to adjust instructions or redirect its actions.
The feature is currently available in limited research preview and is expected to roll out more broadly through Microsoft’s Frontier programme by the end of March 2026 as part of the latest update to Microsoft 365 Copilot.
Microsoft also announced a new Microsoft 365 E7 subscription bundle, combining Office software, Copilot AI and enterprise security tools for businesses.
The company said Copilot Cowork could perform tasks such as rescheduling meetings, generating presentations, conducting deep research on companies and preparing product launch plans.
The announcement comes just days after the U.S. government labelled Anthropic a supply-chain risk amid tensions over the military use of its AI systems.
Microsoft clarified that Anthropic’s technology will remain embedded in its products for clients outside the U.S. Department of Defense, highlighting its strategy of supporting multiple AI models rather than relying on a single provider.
“Copilot hosts the best innovation from across the industry and chooses the right model for the job regardless of who built it,” Microsoft said.
