Claude chat and Cowork merge: what happens to memory and tasks

Quick answer

On September 16, Anthropic announced that chat and Cowork are merging into one Claude experience, beginning with Pro and Max over the coming weeks. This extends the earlier shared-memory update covered here: users no longer choose a mode before starting a task. Below are the migration caveats and the background to shared memory.

Download Chat AI Opens the official App Store or Google Play for your device.

September update: one conversation instead of a mode choice

The unified experience rolls out gradually on web, desktop, and mobile. Accounts that still show Chat and Cowork choices have not migrated yet; there is no switch to enable it. After migration, users cannot return to the separate modes. Anthropic also announced Claude Docs and Claude Slides, with Claude Design inside conversations; all three are beta features on paid plans, with Enterprise activation controlled by admins.

Sources: Anthropic, Anthropic Help Center

What carries over, and what to check after migration

Existing tasks, projects, connectors, skills, artifacts, and files carry over. Older tasks remain accessible in Recents, but search does not include them. Cowork Global instructions move to Settings > General under Instructions for Claude. Memory from computer-only tasks remains attached to those tasks. Manual action approval is the default; local work still needs Claude Desktop open.

Sources: Anthropic Help Center

Background: the August shared-memory update

Claude can now use information remembered from regular chats when a user hands a task to Cowork in the cloud, and context from those Cowork tasks can carry back into chat. Anthropic's documentation describes one shared memory for the two cloud experiences, reducing the need to repeat project details, people, or working preferences during a handoff.

Sources: Anthropic Help Center, TechCrunch

The earlier shared-memory update distinguished cloud and local tasks

The August documentation distinguished shared cloud memory from Cowork sessions running locally on a computer, which did not use that shared memory. Do not treat the September interface merger as evidence that older local-task memory has become shared: the migration documentation explicitly keeps it with those tasks.

Sources: Anthropic Help Center, Anthropic Help Center

Claude saves topics while a conversation is happening

The improved system stores memory as individual topics that Claude can update during a conversation, rather than waiting to create a single summary after the chat ends. Project memories remain separate from one another and from non-project chats. Users can also explicitly ask Claude to remember, change, or forget something.

Sources: Anthropic Help Center, Anthropic Help Center

Memory and past-chat search are different features

Anthropic says memory is on by default for Free, Pro, and Max plans on the web and current desktop and mobile apps. Team and Enterprise owners decide whether their organizations can use it. Searching past chats is separate: Anthropic documents that capability for paid Pro, Max, Team, and Enterprise plans. A user can therefore have memory without having the same past-chat search access.

Sources: Anthropic Help Center

Users can inspect and control what Claude remembers

Saved topics are visible in Settings under Memory, where a user can read, edit, or delete individual items. Pausing memory keeps existing entries but stops Claude from using or adding to them; resetting memory permanently deletes the saved memory and starts again. Incognito chats do not contribute to memory. These controls are different from deleting an ordinary conversation, which does not automatically remove related memory entries in the new system.

Sources: Anthropic Help Center

Sensitive-topic memory requires a separate choice

Anthropic says Claude does not save sensitive topics such as health, race, religion, politics, or gender identity by default. Users may opt in to saving supported sensitive topics, with notices when new items are stored. Some categories, including government ID numbers and financial account numbers, are not saved even after opt-in. Review these settings before using memory for personal or regulated work.

Sources: Anthropic Help Center

Frequently asked questions

What readers usually ask

Does Claude Cowork remember regular Claude chats?

The unified experience uses chat memory for tasks. Memory from older Cowork tasks that ran only on your computer stays with those tasks; the merger does not make it shared.

Is Claude memory available on the Free plan?

Anthropic says memory is on by default for Free, Pro, and Max plans on supported web, desktop, and current mobile apps. Past-chat search is a separate feature documented for paid plans.

Can users edit or delete Claude's memory?

Yes. Users can inspect individual topics in Settings, edit or delete them, pause memory, or reset all saved memory. Resetting is permanent.

Does deleting a Claude conversation delete its memory?

Not automatically in the improved memory system. Anthropic says related memory entries may remain after a conversation is deleted, so users should review and delete the individual memory topics as needed.

Does Claude save sensitive information in memory?

Supported sensitive topics are off by default and require a separate opt-in. Anthropic also lists categories, such as government ID and financial account numbers, that are never saved to memory.

Evidence