Claude now shows how you actually use AI at work

Anthropic launched a beta feature called Reflect in Claude Settings on 9 July 2026. It summarises your chat activity over the past 1, 3, 6, or 12 months, covering key topics, usage patterns, most active day and peak hour, and the types of tasks you run. The dashboard also maps that activity to Anthropic’s 4D AI Fluency Framework: Delegation, Description, Discernment, and Diligence. It offers reflective questions and practical suggestions, such as starting a Project when work needs ongoing context. Users can set quiet hours or break reminders. It is available on Free, Pro, and Max plans for web and desktop when memory is turned on. Insights stay private. Sensitive topics appear only at a high level, and health integrations and incognito chats are excluded. Cowork reflection is still coming later.
Until now, most managers only knew whether AI felt useful in the moment. A good answer in one chat said little about whether the same work could be repeated next week, handed to a colleague, or run without restating the full brief every time. Usage stayed invisible, so weak habits stayed invisible too. Reflect changes that by turning private chat history into a simple operating picture: what you actually ask Claude to do, when the load spikes, and which parts of the work still need stronger briefing, judgment, or human review. The useful signal is not total volume. It is whether your patterns still look like one-off prompting or like the early shape of reusable workflows.
Analysis
Treat this as a diagnostic, not a novelty. Open Reflect for the last one to three months, identify the single recurring task cluster that shows up most often, and convert that one pattern into a Claude Project with standing business context, a clear quality check, and a short note on when the work should stay human.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Claude now shows how you actually use AI at work", Collab365 Spaces. 2 sources referenced.