Anthropic has merged Claude Chat and Cowork into a single product experience. According to the company’s September 16, 2026 announcement, users will no longer need to decide whether a request belongs in a normal conversation or a separate Cowork task. Claude will make that distinction itself.
A short question can still receive a direct answer. A more involved request might turn into a longer-running task in which Claude searches for information, works with files, uses connected services, or produces a finished document. All of that now starts from the same prompt box.
This isn’t a new Claude model, and Anthropic hasn’t attached new benchmark results to the release. It is a significant product change, however. The merger turns the boundary between chatbot and AI agent into an implementation detail that Claude manages on the user’s behalf.
Claude Now Decides How the Work Gets Done
Cowork was originally positioned as the place for “bigger work.” Users could give Claude a more open-ended assignment, allow it to operate across files and tools, and return to the task later. Chat remained the familiar option for questions, writing help, analysis, and shorter interactions.
Anthropic had already begun narrowing that divide. A July 2026 Cowork update brought chat and Cowork into a shared home on web and desktop, while adding cloud-based tasks that could be checked from mobile. The new release removes the remaining mode decision rather than merely placing both options on the same screen.
Under the merged system, Claude decides how much work a prompt requires. It may answer immediately, ask follow-up questions, build an execution plan, or continue operating on the task while the user does something else. Anthropic says people can still steer the process, correct its direction, or take over before the result is finalized.
The announcement does not explain the routing logic behind those decisions. That will be an important part of the experience. Claude needs to recognize when a prompt warrants autonomous work without turning every moderately complicated question into a slow, token-intensive project.
One Conversation Keeps More of the Relevant Context
The strongest part of the merger may be context continuity rather than autonomous execution.
Claude’s migration guide says existing tasks, projects, artifacts, connectors, and skills will transfer into the unified interface. Chats and tasks will appear together in Recents, while memory, preferences, project context, and global instructions can apply across both kinds of work.
Previously, choosing Chat or Cowork could split related work across two spaces. A user might discuss a strategy in Chat, then need to recreate part of that context before asking Cowork to produce the supporting research or documents. The merger gives Claude a better chance of carrying the reasoning, source material, and user preferences into the execution stage.
That also makes the distinction between “conversation” and “task” less useful. A single thread can begin as brainstorming, expand into research, produce a document, and return to a normal discussion for revisions. The user does not need to predict the eventual shape of the work before sending the first message.
Long-Running Tasks Move Into Everyday Claude
The merged experience brings Cowork’s agentic capabilities into normal Claude conversations. According to Anthropic’s Cowork documentation, Claude can search the web, read uploaded files, run code, work with connected services, and create or modify documents through natural-language instructions.
On the desktop app, Claude can also work with local folders, a connected Chrome browser, or the computer more broadly when those permissions are enabled. Users can run multiple tasks at once rather than waiting for one assignment to finish before beginning another.
Cloud-based work can continue after the user closes a laptop. That claim needs a qualification: tasks that depend on local folders, desktop applications, or computer access still require the Claude desktop app and the relevant machine to remain available. Anthropic also supports scheduled and recurring work, although tasks involving local files need the desktop app open at the scheduled time.
The result is a more capable default Claude interface, but not every prompt will use every tool. Access still depends on the user’s plan, device, enabled connectors, permissions, and organizational controls.
Docs, Slides, and Design Become Native Outputs
Anthropic is also bringing three creation formats directly into the merged Claude experience: Docs, Slides, and Design.
Claude Docs turns a conversation into an editable working document. Claude can draft from uploaded files, connected sources, or information already discussed in the thread. Users can edit the document directly, ask Claude for targeted revisions, receive inline comments, and export the result to Google Docs, Microsoft Word, or PDF.
Claude Slides applies a similar workflow to presentations. Users can provide reference material or a brand template, ask Claude to create a deck, and edit individual elements or request broader changes through conversation. Finished presentations can be exported to PowerPoint or PDF. Anthropic says the goal is to produce editable slides rather than static previews.
Sources
- September 16, 2026 announcementclaude.com
- July 2026 Cowork updateclaude.com
- migration guidesupport.claude.com
- Cowork documentationsupport.claude.com
- Claude Designclaude.com
