Anthropic lands Claude inside Microsoft 365

Microsoft 365 is where a huge chunk of office work already happens. That makes distribution the real prize in enterprise AI — not just who has the smartest model in a benchmark. Anthropic’s news is that Claude is now broadly available inside Word, Excel, and PowerPoint, with Outlook opened up in public beta on May 7. The important part is not just “Claude is in Office.” It’s that Anthropic is trying to turn Claude into a work thread that follows you across the apps where the work actually gets finished. ### What actually launched? Anthropic moved Claude for Excel, PowerPoint, and Word from limited access to general availability, and it opened Claude for Outlook in public beta for paid plans. The add-ins are distributed through Microsoft’s marketplace, so this is an in-product workflow, not just a browser tab sitting next to Office. ### Why is Outlook the big piece? (claude.com) Email is where a lot of work starts. A request lands in Outlook, then turns into a draft in Word, numbers in Excel, and a presentation in PowerPoint. Anthropic’s pitch is that the same Claude conversation can carry context across all four apps, so you do not have to keep re-explaining the task every time you switch windows. That is a much bigger idea than a single add-in. ### What can Claude do inside the apps? In Excel, Claude can analyze cells, update assumptions, and help build or revise models. In Word, it can draft and edit text in the document. In PowerPoint, it can help shape slides and presentation content. In Outlook, the beta focuses on email drafting and summarization. Anthropic is basically selling less window-shuffling and less copy-paste glue work. (claude.com) ### Isn’t Microsoft already pushing Copilot? Yes — and that is what makes this interesting. Microsoft has spent the last year turning Microsoft 365 Copilot into a multi-model system instead of a pure OpenAI shop. Anthropic models already show up in Microsoft 365 Copilot products, including Researcher and Copilot Studio, and Microsoft has also rolled out Claude model options in parts of Copilot and Excel. So Claude is arriving through two doors at once: directly as Anthropic-branded add-ins, and indirectly inside Microsoft’s own Copilot stack. (claude.com) ### Why would Microsoft allow that? Because Microsoft now seems to want model choice to be a feature. Its recent Copilot messaging leans hard on “multi-model intelligence,” and Microsoft Learn pages show Anthropic support being managed as part of Microsoft 365 admin controls and subprocessor settings. In plain English, Microsoft wants customers to stay inside Microsoft’s cloud and apps even if the underlying model is not OpenAI’s. (support.microsoft.com) ### What’s the catch? This is not a free-for-all. Access depends on paid Claude plans for Anthropic’s own add-ins, and in Microsoft’s ecosystem some Anthropic features depend on admin settings, geography, and compliance choices. The EU, UK, and EFTA have had extra opt-in requirements in parts of the rollout. So the headline is broad availability, but the fine print is still enterprise software fine print. (microsoft.com) ### Why does this matter beyond Anthropic? Because the fight is moving from model quality to workflow control. The hard part is no longer just generating text. It is being present at the moment work starts, keeping context as work moves, and staying inside the tools companies already pay for. Claude getting deeper into Microsoft 365 means Anthropic is no longer just competing on intelligence — it is competing on placement. (claude.com) ### Bottom line Anthropic did not just ship another Office plug-in. It found a way to make Claude feel more like a cross-app coworker inside Microsoft’s home turf. If that sticks, enterprise AI will look less like “pick one model” and more like “pick the workflow that already owns your day.” (claude.com)