Microsoft today rolled out the biggest agentic upgrade yet to its Copilot platform, introducing Copilot Cowork — a new class of persistent, autonomous agents powered by Anthropic’s Claude models.

These agents go far beyond simple assistance. They can independently handle complex, multi-step workflows across Word, Excel, PowerPoint, Outlook, Teams, and SharePoint — planning, executing, iterating, and delivering complete results with minimal human input. The launch is part of “Wave 3” of Microsoft 365 Copilot and marks the deepest integration yet of third-party frontier models into Microsoft’s core productivity suite.

Key Features Rolling Out Today

• Copilot Cowork agents: Persistent, goal-directed autonomy that maintains context across long-running projects and multiple apps
• Expanded model diversity: Claude (Anthropic) now joins the latest OpenAI models as a first-class option inside mainline Copilot
• New enterprise tiers: Microsoft 365 E7 “Frontier Suite” at $99 per user per month; Agent 365 general availability on May 1 for $15 per user per month

The timing is notable: the announcement came just hours after Anthropic filed its federal lawsuit against the Pentagon, underscoring that commercial demand for Claude remains extremely strong despite the defense controversy.

The Bigger Picture

This is Microsoft’s clearest bet yet that the future of work is agentic. Instead of users prompting tools one step at a time, enterprises will soon have always-on autonomous coworkers that execute entire processes — from financial modeling and contract review to campaign execution and compliance reporting — while staying fully governed inside Microsoft’s security boundary.

By deepening its partnership with Anthropic at the exact moment the company is in open conflict with the Pentagon, Microsoft is sending a strong signal: agentic AI infrastructure is moving into mainstream enterprise production, and commercial adoption is completely decoupled from defense politics.

For organizations tracking the shift to autonomous systems, this is infrastructure-level validation. The tools that will power the next era of agentic workflows are no longer experimental — they are shipping today inside the world’s most widely used productivity platform.