Claude Docs and Slides Explained: What Happened to Claude Cowork?

Claude did not kill Cowork. It removed the reason to choose it.

Anthropic has spent 2026 giving Claude more ways to work. Chat handled ordinary conversations. Cowork handled longer agentic jobs. Claude Code handled software work. Design produced visual artifacts.

On September 16, Anthropic started removing one of those choices.

Claude chat and Claude Cowork are being merged into one Claude. Instead of selecting Chat or Cowork before starting, users describe the outcome and Claude decides which capabilities the task needs. Anthropic says tasks can move from a quick answer into web research, code execution, connected apps, file creation and longer-running work from the same conversation.

Cowork’s capabilities are therefore not being shut down. They are being absorbed into regular Claude.

Claude Code remains separate. Anthropic’s migration guide says the Code tab stays where it is, while Cowork disappears as a mode choice once the new experience reaches an account. 9to5Mac and other launch coverage have described the interface shift as Claude moving from three modes to two: general Claude and Claude Code.

At the same time, Anthropic has launched Claude Docs and Claude Slides, with Claude Design also moving directly inside conversations. Docs and Slides are new beta creation surfaces rather than renamed Cowork features.

The early Choosely read is that this matters more than another feature launch.

Anthropic is simplifying Claude around two broad jobs: let Claude handle general knowledge work from one conversation, or move into Claude Code when the output is software.

Docs and Slides give that first half somewhere for finished work to live.

Cowork is now a capability, not a destination

Before this change, Claude made users understand part of its internal product structure.

A quick conversation belonged in Chat. A larger delegated job could belong in Cowork. Visual work could move toward Design. Starting in the wrong place could mean moving context between surfaces.

Anthropic says users repeatedly found that choice frustrating, so the new interface removes it.

Existing Cowork tasks, projects, connectors, skills, artifacts and files carry over. Cowork tasks appear alongside ordinary conversations in Recents, and the previous Cowork global-instructions setting moves into Claude’s general instructions.

Other controls move too.

The separate web-search toggle disappears and Claude decides when web access helps. Deep Research can be started with /deep-research or through the Research option beneath the input. Older Cowork tasks do not yet appear in Search, although they remain accessible by name in Recents. Memory belonging to Cowork tasks that ran only on the user’s computer remains attached to those tasks.

And once Anthropic migrates an account to the new experience, its current documentation says the user cannot switch back to separate Chat and Cowork choices.

That is a real product migration, not a cosmetic rename.

It also changes the terms of Choosely’s earlier ChatGPT Work vs Claude Cowork comparison, because Cowork is no longer a product a buyer decides to adopt on its own.

Claude Docs is an Artifact with collaboration added

Claude could already produce Word documents, PDFs, spreadsheets and presentation files.

Claude Docs changes the shape of the output.

Anthropic describes a Doc as a rich-text document saved to the user’s Claude account. Every Claude Doc is stored in the Artifacts tab, meaning Docs is not really a standalone document platform sitting beside Claude. It is a richer artifact type inside Claude.

That distinction helps explain the product.

Claude can build a Doc using context it already has from files, memory, Projects, Skills and connected applications. A user can then edit directly, ask Claude for revisions, leave comments or mention @Claude inside a comment to request a change. Multiple editors can work in the same Doc simultaneously.

A Doc can be created from a normal conversation, from an Artifacts template, by starting a request with /docs, or by choosing Output → Docs from the message box. Claude Code can create Docs too: on desktop they open in a side panel, while terminal users receive a link to open the document on the web.

Anthropic’s Docs documentation currently lists these exports:

• Word .docx
• PDF
• Markdown
• Google Docs

A Doc can also be turned directly into a presentation using Claude Slides.

That makes Claude Docs more interesting than “Claude can generate a Word file.”

The document remains part of the AI workflow instead of being the file produced after it.

Claude Slides attacks the next handoff

Claude Slides follows the same logic.

Anthropic says users can create presentations inside a conversation, edit the result, present directly from Claude, share it and export it as PowerPoint or PDF. Docs, Slides and Design all sit within the same broader Claude environment.

The important test is not whether Claude can generate ten attractive slides from a prompt. That is no longer unusual.

The harder problem is whether Claude can take research from several sources, turn it into a coherent report, preserve the important context while converting that report into a presentation, and let a human refine both without repeatedly exporting and re-explaining the job.

That workflow is where the product could become valuable.

It is also where the beta still needs proving.

The unified rollout and Docs access are not the same thing

There are two overlapping release schedules here.

The new unified Chat/Cowork experience is rolling out first to Pro and Max accounts across web, desktop and mobile. Anthropic says more plans will follow, with Enterprise organizations receiving at least 30 days’ notice before their experience changes. Rollout is staged, so two accounts on the same plan may temporarily see different interfaces.

Claude Docs has different availability.

Docs is currently in beta on Pro, Max, Team and Enterprise. It is not available on Free. Anthropic says Docs is enabled by default on Pro, Max and Team; on Enterprise, an Owner or Primary Owner must enable it through Organization settings.

So a Team user can have access to Claude Docs while still seeing the older separate Chat and Cowork experience.

That distinction is easy to miss when the two announcements arrive together.

Claude Docs is not a Google Docs replacement yet

The comparison with Google Docs is unavoidable.

Claude Docs supports live editing, collaboration, comments, sharing and Google Docs export. The Verge describes the launch as narrowing part of Google’s advantage in AI-powered productivity software, while Reuters frames Anthropic’s broader move as part of the contest for enterprise and nontechnical work.

But Anthropic’s own limitations make the current answer straightforward.

Claude Docs is not yet a replacement for Google Docs or Microsoft Word.

There is currently no version history. There is no comment-only permission level. Charts and diagrams do not automatically refresh when underlying connected data changes. Mobile users can view documents but cannot edit them, start from templates or change sharing settings.

And one limitation deserves considerably more attention than it has received in launch coverage:

Deleting a Claude Doc is permanent. There is no trash.

Only the owner can delete a document, but when that happens Anthropic says the Doc becomes unavailable to everyone it had been shared with. There is currently no recovery layer documented.

Pair that with the absence of version history and this is unmistakably beta software.

For consequential documents, those are not cosmetic omissions.

Sharing has another unusual catch

The sharing rules also work differently depending on the subscription.

Pro and Max users can set a Doc so that anyone with the link can open it.

Team and Enterprise users cannot currently share Docs outside their organization.

That sounds surprisingly generous for individual accounts until the next rule arrives:

Opening a shared Claude Doc requires a Claude account.

That creates an obvious adoption wall.

A freelancer cannot simply send a Claude Doc to any client in the way they can send a public Google Docs link and expect frictionless access. The recipient has to enter Anthropic’s ecosystem first.

For collaboration inside an existing group of Claude users, that may be fine.

For documents meant to travel broadly between clients, vendors and organizations, it matters.

Enterprise teams get another reason to wait

Anthropic already excludes some higher-compliance configurations from Claude Docs.

The beta is not currently available to organizations using customer-managed encryption keys, zero data retention or Anthropic’s HIPAA-ready configuration.

There is also a less obvious logging limitation.

Anthropic says events involving the Doc itself appear in the Compliance API activity feed, but activity inside the document, including edits and comments, is not recorded there yet.

For an individual drafting a report, that is probably irrelevant.

For an organization that expects detailed auditability around collaborative editing, it is not.

The feature may be available to an Enterprise organization before it is ready for every Enterprise workflow.

Agentic convenience still spends your Claude allowance

Removing the Chat/Cowork choice makes Claude simpler, but it makes the cost boundary between a simple interaction and a larger agentic job less visible.

Anthropic says everything performed through the unified Claude experience still counts against the account’s usage limits. Longer tasks involving web searches, code execution or file creation generally consume more usage than quick questions. During the staged rollout, Anthropic also warns that usage measurement may differ slightly between accounts on the old and new interfaces.

Choosely’s breakdown of Claude Max limits covers how those session and weekly allowances are structured.

Claude Docs follows the same rule. Larger document jobs, particularly those drawing on several sources, use more of a subscription’s allowance than an ordinary message.

A simpler interface therefore does not make heavy agentic work cheaper.

It just makes the boundary less visible to the user.

The real bet is fewer handoffs

The most useful way to read this release is not “Anthropic built Google Docs.”

It did not.

Claude Docs is an Artifact with collaborative editing. Claude Slides is an AI-native presentation surface. Cowork’s agentic capabilities now begin from ordinary Claude. Claude Code remains the specialist environment for software work.

The pieces are becoming connected enough that a user can begin with a messy problem rather than choosing a product mode.

Research can become a Doc. A Doc can become Slides. Connected apps can provide the source material. Claude can continue larger tasks in the background. Code remains available when the output needs to be software.

That is a much more coherent product than asking the user to decide whether their job is “chat,” “Cowork” or “Design” before they have even started.

The tradeoff is that Anthropic is now asking users to trust Claude with more of the working surface, not merely the answer.

At launch, that surface still has holes.

No version history. No trash. Claude-account-only sharing. Restricted external collaboration on business plans. Incomplete compliance logging.

Those limitations are exactly why this should be treated as an interesting beta rather than a finished office suite.

Choosely verdict

Anthropic has made Claude easier to understand.

Cowork is becoming something Claude does, rather than somewhere users have to go. Docs and Slides then give that agentic work native places to land.

For existing Claude users, that could remove a surprising amount of workflow friction.

But Claude Docs is not ready to displace a mature document platform for important collaborative work. The combination of no version history and irreversible deletion with no trash is enough reason to keep consequential originals somewhere recoverable. Requiring every shared-doc recipient to have a Claude account creates another practical barrier.

The launch is still significant because Anthropic no longer appears satisfied with generating the work and handing it back.

It wants Claude to hold the work while it is being finished.

Whether users trust it with that much of the job will depend on how quickly the boring parts catch up with the impressive ones.

FAQ

Is Claude Cowork discontinued?

Cowork is disappearing as a separate mode for accounts moved onto Anthropic’s new unified Claude experience. Its agentic capabilities remain available from ordinary Claude conversations, and Anthropic says existing Cowork tasks, projects, connectors, skills, artifacts and files carry over.

Can I switch back to separate Chat and Cowork modes?

Not after an account has received the new experience. Anthropic’s current migration guide says migrated accounts cannot switch back to separate Chat and Cowork options.

Is Claude Docs free?

Not currently. Claude Docs is in beta on Pro, Max, Team and Enterprise plans and is not available on Free.

Can Claude Docs replace Google Docs?

Not yet for most serious collaborative workflows. Claude Docs supports editing, comments, collaboration and Google Docs export, but it currently lacks version history and comment-only permissions, requires a Claude account to open a shared Doc and permanently deletes Docs with no trash.

Can anyone open a public Claude Doc link?

No. Pro and Max users can allow anyone with the link to access a Doc, but the recipient still needs a Claude account. Team and Enterprise Docs cannot currently be shared outside the organization.

Can Claude Docs export to Word and Google Docs?

Yes. Anthropic currently supports Word .docx, PDF, Markdown and Google Docs exports.

Can Claude Slides export to PowerPoint?

Yes. Anthropic says Slides can be downloaded as PowerPoint or PDF files.

How do I create a Claude Doc?

Ask Claude for one in a conversation, type /docs, choose Output → Docs, or start from a template in the Artifacts tab. Claude Code can also create a Doc from an active coding session.

Keep your AI stack current

Choosing an AI tool is only the first decision. Pricing changes, features move, plans get restricted and better options appear.

Stack Intelligence watches the tools you already use and surfaces the changes that actually matter to your stack.

Save your tools to Choosely and we’ll keep an eye on what changes, so you do not have to keep checking every pricing page, changelog and product announcement yourself.

Claude changed shape without changing its name. Your other AI tools will too.

Build my AI stack