Claude Cowork finally remembers what you told the app in chat | TechCrunch

Anthropic has introduced a significant update to its Claude artificial intelligence assistant by unifying the memory capabilities of its standard chat function and its collaborative Cowork tool.

Previously, individuals using the platform had to repeatedly brief the AI on details they had already discussed when shifting from the initial brainstorming phase to actual implementation tasks. The newly integrated memory system resolves this common irritation, allowing the virtual assistant to retain contextual knowledge across different workspaces and present itself as a singular, cohesive helper rather than disconnected applications. Additionally, the developer is providing users with a dedicated interface to inspect, revise, or completely delete any saved information to ensure greater personal oversight.

The update also changes how the assistant processes incoming data, capturing and applying fresh details dynamically mid-conversation instead of waiting until a chat session has fully concluded. This real-time processing enables smoother transitions between researching and executing tasks. Addressing potential privacy concerns, the developer has established default safeguards that prevent the storage of sensitive personal information, such as health records, political affiliations, or demographic details, unless users explicitly toggle on a permission setting. Highly restricted data, including government-issued identification numbers, remain permanently excluded from the system's memory to comply with safety guidelines.