Anthropic has merged the memory systems for its Claude chat assistant and its Claude Cowork productivity agent, making user context and history shared across both products. The company announced on August 25 that memory is now two-way, meaning information gathered in one interface is available in the other. This applies to all users on the Free, Pro, and Max plans across web, desktop, and mobile, where the feature is enabled by default. Enterprise and Teams tier users, however, will have both memory and sensitive-topic storage turned off by default, subject to eligibility.
Claude Code, the company’s coding tool, is not included in this integration. The announcement made no mention of that product, and when asked, the company told ZDNET that the update is specifically focused on Claude Cowork and chat. This means conversations or context developed in Claude Code remain separate from the new shared memory system.
The new memory feature allows Claude to use preexisting knowledge from past chat conversations when handling tasks in Cowork. For example, the company said Cowork can draft a manager update based on established preferences, or build a budget and logistics document for a conference using details like headcount, city, and speakers already discussed in chat. Similarly, metrics defined once in chat will automatically appear in future quarterly business review documents created by Cowork without requiring a new briefing.
Users will have several control options. Memory can be paused or reset at any time, and the Settings menu includes new controls under a Memory section. A toggle labeled “Generate memory from chats” controls general recall, while another toggle, “Include sensitive topics in memory,” manages retention of potentially sensitive personal information. The sensitive-topic setting is off by default.
By default, Claude will not store information related to sensitive subject areas, including health, race, ethnicity, religious beliefs, politics, gender identity, and similar categories. Users can grant permission to save such topics, in which case Claude will display a notice indicating that information is being recorded. The company also listed topics it will never store, even with permission, including sensitive identification numbers, criminal history, immigration status, and anything violating its acceptable use policy.
Saved memories are accessible and editable. Claude stores a list of files under Topics in the Memory settings, where users can read, edit, or delete them, and the changes apply globally. The company gave the example that correcting a company’s old name in memory will ensure all future conversations use the new name. This consolidation of features comes as both Anthropic and OpenAI have been moving toward turning their desktop apps into super apps that combine multiple functions.
More AI news from TechManNews.
