Anthropic is launching a new feature in beta that allows users to reflect on and refine how they use Claude by tracking and visualizing their interaction history. The tool aims to help users understand AI integration, assess effectiveness, and align usage with personal goals.
The reflection dashboard, accessible via Settings on the web or desktop app, provides a summary of key topics, usage patterns, and task types over the past one, three, six, or twelve months. It breaks down when users engage most and what they work on, with plans to add time-spent metrics soon.
Users can examine their role with AI using the 4D AI Fluency Framework: Delegation, Description, Discernment, and Diligence. The report offers practical suggestions for improvement, such as starting a Project to avoid re-explaining context.
The feature respects privacy by excluding incognito chats, underlying files from connected tools, and health integration data. It was developed in collaboration with experts from the MIT Media Lab's AHA program, Boston Children’s Hospital, and the Family Online Safety Institute.
The tool is available for Free, Pro, and Max users with memory enabled. Reflecting on Cowork conversations will be added soon.