Claude now carries one memory from chat into work tasks
Reviewed by Helen Jones
On 25 August 2026 Anthropic merged Claude's chat memory with Claude Cowork, so notes taken in a conversation can travel into a cloud task and findings can come back the other way. Memory now updates while you talk, not only after the chat ends. Saved items show up as topic files you can read, edit, or delete. Memory is on by default for Free, Pro, and Max on web, desktop, and mobile. Team and Enterprise plans keep it off for individuals until an admin turns it on. Claude Code is not included. Health, race, ethnicity, religion, politics, and gender identity are not stored by default. Some data, including government IDs, is never stored. Sensitive-topic storage is not retroactive, and phones need the latest app.
Until now, a useful Claude session died when the window closed. You restated audience, standards, and business context every time, then hoped Cowork would not start from a blank slate. That is why AI felt like a slot machine even when the first draft looked fast. Shared, live memory changes the failure mode. Wrong facts and half-true preferences can now follow you into later chats and into Cowork. The gain is a reusable context library. The cost is that someone has to treat those topic files like a briefing pack with a human review step, not like a private diary the model is allowed to keep forever.
Analysis
This is a change to act on, not a trend to watch. Open Claude's memory files this week, delete anything you would not put in a shared briefing, and write one short topic file for a recurring task with the quality check you still want a human to run.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Claude now carries one memory from chat into work tasks", Collab365 Spaces. 3 sources referenced.