Anthropic Just Deleted the Line Between Chatting and Working
Cowork is gone as a separate thing. As of September 16 Anthropic folded it back into Claude, so there is one product again: you type, and Claude decides on its own whether this is a question to answer or a job to go do. No mode switch, no second app, no "open the agent surface." The blog post is at https://claude.com/blog/cowork-is-now-claude and the rollout starts with Pro and Max on web, desktop and mobile, with Team and Free following, and Enterprise getting 30 days of notice before anything moves.
The thing that came along with it matters more than the merge. Claude Docs and Slides are now in beta on paid plans, which means the output of a long agent run is a document or a deck you can open at a link or export to PowerPoint and PDF, not a wall of chat you have to copy out by hand. Design also moved into the conversation. Skills, connectors and history follow you across all of it instead of living in whichever surface you happened to start in.
Worth remembering how recent the split was. Cowork shipped as a separate workspace because the industry consensus said chat and agentic work were different products with different interfaces. A few months of real usage later, the separate workspace is the part that got deleted. Every vendor that just spent a quarter building a distinct agent canvas next to their chat box should read that carefully.
The one genuinely new control is the ask-first setting: you choose whether Claude checks with you before acting or only checks in when it needs to. That is the whole autonomy debate reduced to a toggle, and putting it in the consumer UI is a bigger statement than Docs and Slides combined. It also means the permission question moves from "which app am I in" to "how much rope did I give it," which is harder to reason about and much harder to undo.
Related reading: [the three labs are already negotiating who gets to certify them](https://clauday.com/article/c87f9d35-ee9d-48dd-b2ac-856d5d3adedc)
← Back to all articles
The thing that came along with it matters more than the merge. Claude Docs and Slides are now in beta on paid plans, which means the output of a long agent run is a document or a deck you can open at a link or export to PowerPoint and PDF, not a wall of chat you have to copy out by hand. Design also moved into the conversation. Skills, connectors and history follow you across all of it instead of living in whichever surface you happened to start in.
Worth remembering how recent the split was. Cowork shipped as a separate workspace because the industry consensus said chat and agentic work were different products with different interfaces. A few months of real usage later, the separate workspace is the part that got deleted. Every vendor that just spent a quarter building a distinct agent canvas next to their chat box should read that carefully.
The one genuinely new control is the ask-first setting: you choose whether Claude checks with you before acting or only checks in when it needs to. That is the whole autonomy debate reduced to a toggle, and putting it in the consumer UI is a bigger statement than Docs and Slides combined. It also means the permission question moves from "which app am I in" to "how much rope did I give it," which is harder to reason about and much harder to undo.
Related reading: [the three labs are already negotiating who gets to certify them](https://clauday.com/article/c87f9d35-ee9d-48dd-b2ac-856d5d3adedc)
Comments