Microsoft is embedding Anthropic’s Claude AI into its Copilot platform, marking a major shift in the company’s enterprise AI strategy.

The integration powers a new feature called Copilot Cowork, designed to turn Copilot from a conversational assistant into an autonomous digital coworker capable of executing multistep tasks across Microsoft 365 applications, according to Microsoft’s release today.

Copilot Cowork uses Claude’s agentic capabilities to plan and carry out complex workflows instead of merely generating text or answering questions. The agentic AI can take action on behalf of users building presentations, analyzing data in spreadsheets, drafting emails and coordinating meetings automatically, the release stated.

Claude’s role inside Copilot is to provide the reasoning and planning layer required for these “AI agents.” Anthropic’s models are widely regarded as strong at long-context reasoning and structured workflows, making them well suited for tasks that require multiple steps or interaction across files and apps.

By embedding Claude into Copilot, Microsoft aims to enable autonomous work scenarios such as compiling financial reports from scattered documents, generating slide decks from research files or managing team calendars with minimal human input, according to the release.

Copilot Cowork runs within Microsoft 365’s existing security and governance framework, meaning AI actions follow organizational permissions and compliance policies by default.

Strategically, the move also signals Microsoft’s shift toward a multimodel AI approach. While Copilot has historically relied heavily on OpenAI models, the integration of Anthropic’s Claude allows Microsoft to select the best model for specific tasks and reduces its dependence on a single AI partner.

Copilot Cowork is initially launching as a limited preview through Microsoft’s Frontier program, with broader availability expected later in 2026.

Read more coverage from FinAi Banking Summit here.