Anthropic has started a beta test of Reflection, a Claude feature that gives people a recap of how they have used the AI assistant over time. The new view is designed to show broad patterns—such as recurring topics, common task types and periods of activity—rather than simply encouraging more time in the app.
Reflection is available in Claude’s settings for eligible Free, Pro and Max users on the web and desktop when memory is turned on. The company is also adding optional quiet hours and break reminders, extending the idea from a retrospective dashboard to a set of controls for when Claude should stay out of the way.
The settings are optional. A person can use the reminder controls to set boundaries around when the assistant can interrupt, or leave them off and use the recap only as an occasional check-in.
What the feature shows
Users can choose a one-, three-, six- or 12-month lookback. The recap can identify themes in the conversations a person has had with Claude, the sorts of work they delegate and patterns such as their busiest day or peak hour.
The point is not to turn every interaction into a productivity score. Anthropic frames the tool as a chance to ask whether using AI is helping with a person’s goals, and whether there are tasks they would rather continue doing themselves.
What is—and is not—in the recap
The feature depends on Claude memory, so it is not a universal history of every chat. Anthropic says Reflection does not pull the contents of files in connected tools, does not use incognito chats and does not reference conversations connected to a health integration. Sensitive subjects can still appear in a high-level summary, which is an important distinction for anyone deciding whether to enable memory.
That makes the privacy setting central to the product. A useful recap depends on enough retained context to spot habits; the same context can reveal something about a person’s routines, work and interests. Users who want the recap should review their memory settings before treating it as a neutral activity log.
Why this is a notable AI experiment
Chatbots increasingly serve as writing aid, research tool, planner and work companion. Reflection applies a familiar “year in review” idea to that relationship, but with a different question: not only what did you do with the tool, but when might you want to use it less?
That is a meaningful product choice as AI assistants become more embedded in daily workflows. The new controls do not settle the broader questions around reliance or data handling, but they give users a visible place to evaluate their own patterns rather than leaving that work entirely to the platform.
What to watch next
Reflection is in beta, and Anthropic says a version is planned for Claude Cowork. The practical test will be whether people find the summaries genuinely useful without feeling that a personal AI dashboard exposes more about their lives than they expected.
