Anthropic announced on 16 September that Claude Cowork — the separate workspace the company built for bigger, longer jobs — is being merged into the ordinary Claude chat window. There is no longer a decision to make about which tab a task belongs in. The model routes it.
Two new tools arrived in the same announcement. Claude Docs and Claude Slides let a chat produce an editable document or a presentation that exports to PowerPoint or PDF, shareable on a single link that opens on a phone. Claude Design, which launched in April for building websites and prototypes, now runs inside any conversation too.
Why the split existed, and why it died
Anthropic’s own post is unusually candid about the reason: Cowork was built as “a separate place for bigger work,” Design for visual work, and the complaint from people using both was that the frustrating part was deciding where a task belonged — and that work started in one didn’t carry into the other. The fix is “one Claude” where, in the company’s words, “you don’t have to choose where a task goes. Claude does more of the work.”
In practice this is a routing problem. Ask for a quick answer and you get chat. Hand over a report due at noon and the same window spins up a background task, checks in from your phone, and has the doc and matching slides waiting when you sit down at your desk. Users who want to keep control can leave Claude asking before every action; the ones who don’t can let it run and only get pinged when something needs a closer look.
The Google gap it’s meant to close
The competitive read is the more interesting half. TechCrunch notes the rollout lands alongside dedicated document and presentation features that put Anthropic head-to-head with Google’s productivity suite — Docs windows, editable files, shareable links. The Verge says it plainly: Claude’s models are generally rated ahead of Google’s, but Google’s AI is natively wired into the tools people already work in, and this update “starts to close the gap between Anthropic’s AI platform and Google’s productivity tools, giving users fewer reasons to leave a Claude chat.”
That is the same bet OpenAI made in July when it turned ChatGPT into an agent that works across connected apps rather than just answering in a chat box — we covered that shift here. The frontier is no longer just model quality; it is whether the model sits where the work already happens.
Rollout and caveats
The unified experience is rolling out to Pro and Max subscribers on web, desktop and mobile “over the next few weeks,” with Team and Free plans to follow and Enterprise admins getting at least 30 days’ notice. Anthropic says nothing needs to be turned on, chat-only users should notice no difference, and everything in Cowork — projects, artifacts, connectors, skills — carries over. Docs and Slides are beta on paid plans.
One thing the merger doesn’t fix: the sandbox. In July, security researchers showed a multi-stage exploit chain that escalated to root inside Cowork’s Windows sandbox, and Anthropic’s classification of it as “not a security issue” drew criticism — we covered that finding here. Folding Cowork into the main chat means that attack surface now belongs to everyone’s default Claude, which makes the containment engineering matter more, not less.
The NZ takeaway
For New Zealand’s small-business and public-sector users watching the agent-tools race, the significance is less the interface and more the direction of travel: assistant vendors are collapsing “chat” and “does the work” into one product, and pricing the document and slide outputs as part of the subscription. Google moved Gemini deeper into Docs and Sheets earlier this year; Anthropic’s answer is to rebuild enough of a workspace that you don’t need to open one. The next differentiator is which assistant can be trusted to run unattended — and that is a safety question as much as a feature list.
