A change to Claude’s memory system will reduce one of the most frustrating parts of using AI agents — the constant need to rebrief the AI on information it already knows.
Anthropic announced on Tuesday that it is merging the memory system used by Claude’s chat and Claude Cowork. This means Claude will be able to remember what it learned in one area even when users move to another.
Anthropic is also giving users more control over the information Claude retains. Users will be able to view, edit or delete stored information on any topic.
The update is expected to make it easier for users to move between the research stage of a project and actually taking action.
Previously, users could spend time developing ideas and plans through conversations with Claude. When they later moved to Cowork to carry out the work, they often had to start again by explaining the project, its background and other important details.
This could make the experience feel fragmented and frustrating. The new system is intended to make Claude feel more like one continuous assistant rather than two separate products operating under the same platform.
Anthropic gave the example of someone preparing an update for their manager or creating an agenda for a conference. Cowork could already understand references from earlier conversations about the project.
For example, it could know the conference headcount, city and speakers without requiring the user to enter those details again.
The shared memory system should therefore allow users to move more smoothly from discussing an idea with Claude to using Cowork to complete the related work, without repeatedly providing the same context.
