Claude for Word brings tracked-change editing into Microsoft documents

What Claude launched

On April 10, 2026, Claude announced that Claude for Word is now in beta. The product puts Claude in a sidebar inside Microsoft Word so users can draft, edit, and revise documents without moving the text into a separate chat window. The announcement and product page both emphasize that edits appear as tracked changes, which matters because enterprise document work is usually constrained by legal review, executive approval, and auditability rather than pure text generation speed.

That positioning makes this more than another writing plug-in. Anthropic is trying to land Claude directly inside a document workflow that organizations already trust, instead of asking teams to copy text back and forth between Word and a browser. In practice, that lowers the friction for using AI on contracts, board materials, policy docs, sales collateral, and internal memos.

What the beta includes

The Claude for Word page says the beta is available for Claude Team and Enterprise plans. It supports .docx and .docm files, preserves existing heading styles, numbering, bullet formatting, and defined terms, and can work with comments by editing anchored text as tracked changes and replying in the thread with what it changed.

• Draft or rewrite text directly from the Word sidebar.
• Keep existing document formatting intact instead of flattening the file into plain text.
• Return edits in tracked changes so reviewers can accept or reject them.
• Operate within the organization’s existing security and compliance framework, according to the product page.

Why this matters

The key signal is workflow integration. Many AI writing tools can generate prose, but enterprise teams still live inside Word because approval chains, redlines, and document ownership already run through it. By preserving formatting and using tracked changes, Claude is adapting to the review mechanics that govern high-stakes documents rather than asking teams to abandon them.

An inference from the launch is that Anthropic sees document-native assistants as a serious expansion path beyond chat and coding. If an assistant can work directly inside the tools where teams negotiate, annotate, and approve work, it becomes harder to treat as an optional side tool. The constraint is equally clear: Anthropic itself says users should review the tracked changes before accepting them. That caveat is correct. In this category, success depends less on raw fluency than on whether the assistant stays useful while remaining visible, reviewable, and easy to govern.

Sources: Claude X post · Claude for Word page