Anthropic merges Claude Cowork and chat into one Claude
Hand over a task, close the laptop, and Claude keeps going, with Docs, Slides and Design now inside every conversation.
Anthropic says Cowork and chat are merging into one Claude. The pitch is that you can ask a quick question or hand over a report in the same place, and Claude carries on with the work after you close your laptop. If something is unclear, it asks, and the account is explicit that you keep the final say. It rolls out to Pro and Max over the next few weeks.
Ask a quick question or hand over a report, and Claude takes it from there, even after you close your laptop.
Boris Cherny added the second half of the announcement. Claude Docs, Claude Slides and Claude Design are now in every conversation. Ask for a presentation and you get one you can open, edit and export as PowerPoint or PDF, and the same goes for a document or a design. There is no separate tool to open, they sit in the chat.
Dan Shipper's reaction was three words long and captured why people care.
What is actually new here
Two things, and they are different in kind. The artifact part is a consolidation. Slides, docs and design were things you went somewhere else for, and now they are produced inside the thread you were already in, in formats a client will accept without asking questions.
The unattended part is the bigger shift. A chat assistant is something you sit in front of. A task that survives you closing the laptop is something closer to a worker that checks in when it hits an ambiguity. That changes the unit of work from a message to a job, which is the thing founders have been building agent rigs to get for themselves.
The open question is cost
Anthropic did not publish usage numbers, quality numbers or any indication of what a long unattended run consumes. For anyone on Pro rather than Max, that is the whole question. A model that keeps working while you are away is only useful if you can afford to let it, and nobody outside Anthropic knows yet where the limits land.
The second unknown is how often it stops to ask. An agent that interrupts constantly is a chat window with extra steps. One that never interrupts is a agent that confidently produces the wrong report while you are at dinner. The announcement promises the middle path, and the middle path is the hard part.
For a solo founder the practical read is to wait for the rollout and test it on work where a wrong answer is cheap. Decks, first drafts, research summaries. The claim that it keeps going after you close the laptop is Anthropic's claim, not a measured result, and the people we follow have not run it long enough to say otherwise.
THE END OF THE TABS ERA IS UPON US