You no longer pick a mode in Claude: chat, agent work, docs and slides in one interface
On September 16, 2026 Anthropic announced Claude Cowork and chat are merging into one Claude experience. Users no longer decide where a task goes; Claude works out what it needs. The same post moved Claude Design inside conversations and introduced Claude Docs and Claude Slides. All three are in beta on paid plans, rolling out to Pro and Max first.
Three modes become two: Cowork moves into chat, Claude Code stays separate
According to Anthropic's official blog post, Claude Cowork and Claude chat are now one Claude. The user does not choose where to take a task; Claude determines what the task needs. The rationale, as 9to5Mac relays it from Anthropic's post, is that people had trouble deciding which mode or tool a task belonged in, and work started in one tool did not carry over to the other; TechCrunch sums it up as customers struggling to pick the right tab for the right task. 9to5Mac notes that Claude's modes go from three to two: chat and Claude Code. Claude Code remains a separate product; SiliconANGLE writes that it is unaffected by the change for the time being, while Anthropic's post only mentions that Claude Code now supports artifacts. TechCrunch adds that Artifacts are also folded into the single interface. On behavior, Anthropic is clear: Claude asks before taking actions by default and can be set to work independently. One example in the post is scheduling a task to repeat every Monday.
Claude Design moves into conversations, Docs and Slides live at one link
The new package has three parts. Claude Design, introduced in April for website and prototype design according to TechCrunch, now works anywhere inside Claude. With Claude Slides, users ask Claude to create, edit and present a deck; per Anthropic it can be presented straight from Claude or downloaded as PowerPoint or PDF. In Claude Docs, per TechCrunch, the user works alongside Claude section by section, asks questions and comments on completed parts. Anthropic says anything made with Design, Slides or Docs lives at one shareable link that opens on a phone, and users can leave a comment for Claude on a slide. Export of documents to Microsoft Word or Google Docs is reported only by SiliconANGLE, not by Anthropic or TechCrunch. On access, Anthropic's wording is precise: "All three are in beta on paid plans," and Enterprise admins choose when to turn them on. The rollout starts on Pro and Max plans over the coming weeks, on web, desktop and mobile; Team and Free plans follow soon, with no dates given. Enterprise admins get at least 30 days' notice before anything changes, and 9to5Mac adds that users do not need to enable anything manually.
The chat window becomes a workspace, but the cost question stays open
For a business user the core change is simple: the question of which mode to use disappears. Until now a task meant choosing between chat and Cowork, and as the coverage relays, work started on one side did not travel to the other. Now questions, agent work, documents and slides happen in the same window, and the output sits at a single link that opens on a phone, with room for a comment to Claude on a slide. The assistant moves from being an answer box towards being a shared working surface. Asking for approval before acting by default is another detail that matters at work: whoever wants automation switches to independent mode, whoever is cautious keeps the approval gate. One reservation comes from SiliconANGLE's Mike Wheatley as an editorial note, not from Anthropic: tool-calling agentic flows consume far more tokens than a simple chat exchange, so if simple questions are routed through Cowork's agentic machinery, users could push up compute cost without realizing it. The announcement mentions no pricing change and gives no date for when the rollout will be complete.
For SMBs in Türkiye the tool-learning burden shrinks, while access remains an assumption
First the boundary: the announcement gives no country or regional availability and does not mention Türkiye. Since the features are tied to paid plans, existing Pro and Max users in Türkiye can reasonably expect the same timeline, but that is an assumption; Anthropic did not say it. For a small team the real gain is in the number of tools. In an SMB the same person often writes the proposal, prepares the client deck and follows up on recurring tasks; teaching a separate tool for each and moving files between them costs time. Bringing chat, agent work, documents and slides together removes that burden and takes the mode question off the table. It is still a beta, though: no date was given for the Team plan, so companies on it will wait, and in Enterprise the switch sits with the admin. Because the agent can be set to work independently, a business handling customer data should define up front who runs what, with which permissions. SiliconANGLE's token warning applies to small budgets too: keeping simple tasks out of the agentic flow is a habit worth setting from day one.
The UNALSOFT view
Our reading is this: the major tools are moving toward a single interface, which shows the choice is shifting from the tool to the process. In the agentic AI solutions we build for clients, the focus is not which tool to pick but which task runs at which step, behind which approval gate and on which data. Claude deciding by task rather than by mode confirms that approach. As interfaces get simpler, the gap between a well-designed process and scattered usage becomes more visible. We do not know the timeline for Türkiye; we do know that process design does the same job in every tool.
Sources
Anthropic, "Claude Cowork and chat are now one Claude" · TechCrunch report · 9to5Mac report · SiliconANGLE report
Want to tie your team's AI usage to a single process?
Let's design together which tasks run in chat, which run through an agent, and where the approval gates and data boundaries sit. A short conversation is enough to start.