The new option is intended for situations where someone needs to return to a task or review past research without going over old ground. When prompted, Claude searches earlier chats linked to the same workspace and project. It then returns a summary and offers to continue the work. The feature does not run automatically and can be switched off in the settings menu.
Anthropic’s design keeps the function limited in scope compared with memory systems from other AI providers. OpenAI’s ChatGPT, for example, stores all conversation history to shape later responses. Google’s Gemini can also draw on earlier discussions and, in some cases, search history. Claude’s feature instead focuses on retrieval at the user’s request, avoiding a constant record.
In a demonstration, Anthropic showed Claude reviewing a series of exchanges from before a user’s holiday, summarizing the progress, and offering to pick up the same project. The approach aims to keep context available while maintaining tighter control over stored information.
The rollout has started for eligible plans, with additional tiers expected to gain access in the coming months.
Notes: This post was edited/created using GenAI tools.
Read next: Mobile Devices Cement Lead in Global Web Traffic as Desktop Use Declines