Anthropic folds Claude Cowork into chat and adds Docs and Slides

Anthropic has folded Claude Cowork into Claude's main chat and launched two new tools beside it, Claude Docs and Claude Slides. The change was announced on 16 September and it removes a decision the company says its users never wanted to make: which surface a piece of work belonged on.

What actually changed

Cowork, the agentic workspace Anthropic built for longer jobs, no longer sits apart from chat. What it can do is now reachable from any conversation, arriving with the context, skills and connectors a user already has in place. Claude Design, previously its own space for visual work, moves inside the conversation too, while anyone using it standalone keeps the old behaviour. Docs and Slides are new: ask for a document and you and Claude write it together, ask for a presentation and Claude drafts the slides, which can be edited in place, presented straight from Claude, or exported as PowerPoint or PDF.

Who gets it, and when

All three are in beta on paid plans. Anthropic is starting with Pro and Max and spreading the rollout over the next few weeks, with Team and Free to follow. There is nothing to switch on. Enterprise administrators decide when the tools are turned on for their organisations, and Anthropic has promised them at least 30 days' notice before anything changes. Existing Cowork users keep their chats, projects, artifacts, connectors and skills, which is the detail that matters most to anyone who has already built a working routine around it.

One conversation, one link

Anthropic's own example is a weekly report due at noon: ask what moved in the pipeline, then ask for it written the usual way, with anything that slipped flagged and the highlights in five slides for the leadership meeting. Because both come out of the same conversation, the deck already matches the document. Anything made with Docs, Slides or Design lives at a single shareable link that opens on a phone, and an element can be moved by selecting it or by telling Claude what to change. By default Claude asks before taking an action, and a setting lets it keep working and check in only when something needs a closer look.

The company's customer quote for the launch comes from Andrew Keller, a senior economist, who describes Claude pulling up a legal research database, reading the cases, working out which further cases might be needed, downloading them and storing them in a folder for personal review.

The file, not the model, is the battleground

For two years the assistant was the product and the model was the pitch. Merging Cowork, Design, Docs and Slides into one conversation says something different: the output is the product, and the place a person opens is just where the work starts. That puts Claude against Word and PowerPoint far more directly than against another chatbot, because the switching cost in offices sits in the file format and the habit, not in who writes the best paragraph.

Our opinion

Anthropic is making the right call on the boring parts. Nothing to enable, existing Cowork work preserved, administrators given 30 days' notice, and Claude still asking permission before it acts unless you tell it otherwise. That is a deliberate contrast with the feature-every-Tuesday posture of its rivals, and it is the reason enterprises will actually test this rather than block it. The catch is concentration. When a single conversation produces the report, the slides and the folder the research landed in, a bad answer no longer costs you one paragraph, it propagates through everything downstream, and the shareable link makes that easy to circulate. Concise beats sprawling for getting work out the door, but the merged Claude only earns the trust Anthropic is asking for if its default of checking in survives the pressure to look faster than everyone else.