Anthropic launched the Reflect feature test version, turning Claude into a mirror for users to review AI usage habits. Available to free, Pro, and Max users with memory enabled. Users can analyze past 1-12 months of chats, receiving a personalized report on key topics, usage patterns, and active periods. The report evaluates four dimensions: task delegation, goal description, result identification, and responsible caution, offering improvement suggestions. Privacy is protected—no incognito or health data is read. Future updates will include visual timelines and silent period settings.