Anthropic Introduces Memory And Incognito Mode To Claude AI

Anthropic, the company behind Claude AI, has rolled out a new “memory” feature designed to help users streamline work by allowing the chatbot to retain context across projects and conversations.

The feature, launched this week for Team and Enterprise plan users, enables Claude to remember project details, client needs, and team workflows, eliminating the need to re-explain instructions. Anthropic says this functionality can help sales teams track client interactions, product teams manage specifications, and executives monitor initiatives more efficiently.

Memory is optional, with controls that let users view, edit, or delete stored information. Project-specific memories keep unrelated tasks separate, ensuring that client work, product launches, and confidential discussions remain distinct. A memory summary page allows users to monitor exactly what the AI retains.

Alongside memory, Anthropic has introduced an “Incognito chat” mode available to all users. The feature allows conversations that do not save to history or memory, designed for sensitive brainstorming, private strategy sessions, or instances where a clean slate is needed.

Enterprise administrators can disable memory for their organizations at any time. Regular data retention policies remain in effect for Team and Enterprise plans.

Anthropic said it is taking a phased approach to deploying memory, emphasizing safeguards to avoid inappropriate use in sensitive contexts. The company stressed that the tool was built with workplace settings in mind and will continue to be tested before wider expansion.

To activate memory, users can enable it in settings, where Claude automatically generates a record based on past chats. They can also transfer data from other AI tools or export Claude’s memory for backup.

“Great work builds over time,” Anthropic said in its announcement. “With memory, each conversation with Claude improves the next.”