Claude rolled out a memory feature this week, but it is limited to select paid users of the AI model. Just days after Anthropic disclosed that Claude can create spreadsheets and presentations, they launched a memory capability for Team and Enterprise plan users. According to a blog post published on Thursday, the AI model now “remembers you and your team’s projects and preferences.”

The memory functionality of Claude is not accessible to free users or those subscribing to the individual Pro or Max plans. However, Anthropic also revealed the introduction of incognito chats (similar to the incognito window in Chrome), which are available for all users. These chats do not show up in a user’s conversation history or get stored in memory.

For users utilizing Claude for work, the model can now “remember” professional environments, encompassing client requirements and project specifics, together with your team’s habits, such as your workflows. Additionally, Claude will establish a separate memory for each project if you utilize the projects feature (a “self-contained” workspace with its own chat relevant to that topic).

According to Anthropic, memory incorporates “new safety considerations” aimed at facilitating work while preventing the storage of sensitive conversations and subjects, although the blog post does not provide further details. Anthropic plans to assess and test the functionality of memory before widening its accessibility to additional users.

For those on Team and Enterprise plans, Claude’s memory can be activated in the app’s settings. As stated in the blog post, users can view precisely what Claude retains and modify “memory summaries” to emphasize or disregard certain details. Enterprise administrators can also deactivate memory for all users under their plan at any point.

Users have the option to import memory from other AI tools into Claude or to export their Claude memory. Considering that Google Gemini and ChatGPT have had memory capabilities for some time — and have even enhanced the feature — this represents a move towards Anthropic catching up.

Disclosure: Ziff Davis, the parent company of Mashable, filed a lawsuit against OpenAI in April, claiming it violated Ziff Davis copyrights in the training and operation of its AI systems.