Claude can now reference past chats, if you want it to

Claude is getting a better, if selective, memory. Rather than acting as perfect catalog of everything you’ve talked about or shared, Anthropic says the AI chatbot now has the ability to reference past chats when asked, so you don’t have to re-explain yourself.

The feature seems like it could help you pick up a work project after time away, or query Claude for the details of a past research session that you don’t quite remember. The key point is that Claude has to be prompted: It doesn’t call on past chats unless you specifically ask it to. Anthropic also says that Claude’s ability to reference chats is specific to the workspace and project you’re working in.

Anthropic’s implementation is much more limited in scope compared to how ChatGPT’s memory feature works. OpenAI updated ChatGPT in April to save all conversations you have with it, and rely on those records to personalize answers to any new question or prompt you send. The basic idea is that every conversation could improve the chatbot’s responses. Google lets Gemini similarly recall past conversations and has also tested using Google Search history to further personalize AI responses.

In contrast to both, Claude is really only performing a search of past conversations when asked, not referencing a profile of past chats. It’s more privacy-minded by default and you can disable Claude’s ability to do it at all via a settings toggle.

If you’re subscribed to the Max, Team or Enterprise plans, Claude’s new ability should be rolling out now, according to Anthropic. The company says the feature will expand to other plans soon.