Anthropic's Claude chatbot and its workplace assistant Cowork now share a unified memory system, meaning everything one service learns about users is automatically available to the other, the company announced Tuesday. The frontier AI lab confirmed that users have no option to separate the two services' memories because they operate on a shared backend. The integration aims to help Cowork automate office tasks like writing reports, building presentations, and compiling budget documents by drawing on information users previously shared with Claude Chat.
The shared memory works in both directions: what Claude learns through conversations is accessible to Cowork, and what Cowork picks up while automating tasks flows back to Claude. Both services can store details about users' interests, the names of friends, family members, and colleagues, or nearly anything else mentioned during interactions. Memory generation now happens in real time during chats rather than through post-conversation review, as was previously the case. The feature is turned on by default for users on free, Pro, and Max subscription tiers, though Cowork itself isn't available to free-tier accounts. Claude Code's memory remains isolated from the shared system, and Anthropic told The Register it has nothing to announce about whether that will change.
According to Anthropic, the integration is particularly useful for workplace automation, with the company offering examples like asking Cowork to draft a manager update when it already knows "who that is and how she likes updates written." The firm also suggested brainstorming a conference agenda with Claude Chat and then relying on Cowork to remember event details while building budget and logistics files, or defining metric presentation formats once in Claude and trusting Cowork to recall that structure every quarter. The system doesn't remember sensitive information related to health, ethnicity, religious beliefs, politics, or gender identity by default, though users can opt in to storing those topics if they choose.
Users who want to keep Claude and Cowork's knowledge separate must create different accounts for each service, an Anthropic spokesperson confirmed. Other privacy controls include pausing memory creation entirely, using incognito mode for conversations users don't want stored, or manually deleting individual memories from Claude's knowledge base. For Team and Enterprise customers, administrators control whether the memory feature is enabled at all, meaning individual employees will need IT approval to activate it. The shift reflects a broader industry trend toward AI systems that accumulate persistent user context rather than treating each interaction as isolated, though it raises questions about how much control users have over where their information flows once they share it with one tool in an ecosystem. As these memory systems grow more interconnected across workplace productivity suites, the line between optional convenience and mandatory data-sharing may become harder for organizations to draw without fragmenting their tech stack.