Jonathon's AI Wiki

Reflect with Claude — Usage-Reflection Dashboard (Beta)

4 min read

Source: raw/x-account-claudeai-2075225957711929667.md, ai-research/reflect-with-claude-anthropic-2026-07-09.md

Reflect with Claude is a beta feature Anthropic introduced on 2026-07-09 that lets you track and visualize how you use Claude, then decide whether that time aligns with your goals. It lives in Settings on Claude for web or the desktop app and is framed around mindful, healthy AI use — the product of collaboration with digital-wellbeing researchers. It runs only for users who have Memory enabled and deliberately excludes incognito chats, connected-tool source files, and any health-integration conversations.

Key Takeaways

  • What it is: a usage-reflection dashboard (beta) in Settings on Claude web/desktop, launched 2026-07-09 for Free, Pro, and Max users who have Memory turned on.
  • What it shows: a summary of your key topics, usage patterns, and task types; a breakdown of when you use Claude most and what you worked on; and a lookback window of 1, 3, 6, or 12 months. A view of how much time you have spent using Claude is “coming soon.”
  • Reflective prompts: it periodically surfaces questions like “What’s one thing you want to keep doing yourself, even if Claude could do it faster?” and lets you talk them through with Claude.
  • Wellbeing controls: you can set quiet hours or schedule a nudge to take a break after a certain amount of usage. Both are dismissible reminders of your own stated preferences, not hard usage limits.
  • 4D AI Fluency Framework: activity is summarized across Delegation, Description, Discernment, and Diligence, with examples of how you collaborate (e.g., reworking email drafts in your own voice, or delegating only after settling strategy yourself) and practical suggestions (e.g., start a Project instead of re-explaining context each time).
  • Privacy by design: insights don’t draw from incognito chats and don’t pull the underlying files from connected tools — a summarized inbox may appear in your reflection, but the source emails won’t — and health-integration conversations are excluded entirely. Sensitive conversations can still appear, but only at a high level, and the insights aren’t reused for any other purpose.
  • Built with wellbeing experts: developed with the MIT Media Lab’s Advancing Humans with AI (AHA) program, the Digital Wellness Lab at Boston Children’s Hospital, and the Family Online Safety Institute.
  • Memory-gated: if a report won’t generate, the usual cause is Memory being off. Reflecting on Cowork conversations is coming soon.

Try It

  • Turn Memory on (Settings > Capabilities > Memory), then open Settings on Claude web or desktop and select the option to reflect on your usage to generate your report.
  • Pick a lookback window (1, 3, 6, or 12 months) and read the when/what breakdown — check whether your peak-usage times and top topics match your intended priorities.
  • If the reflection shows usage bleeding into time you’d rather protect, set quiet hours or a break nudge.
  • Use the 4D AI Fluency breakdown as a self-audit: for each dimension (Delegation, Description, Discernment, Diligence), note one habit to keep and one to change; act on a suggestion such as converting a recurring context-heavy chat into a Project.
  • Treat it as a periodic check-in — revisit monthly rather than reading it once.

Open Questions

  • No time-spent metric yet — Anthropic says a view of total time using Claude is “coming soon” but gives no date.
  • Cowork reflection is not yet available (also “coming soon”).
  • Enterprise/Team availability is unstated — the beta is named for Free, Pro, and Max only.
  • How the insights are generated (which model, on-device vs. server, retention window) isn’t detailed beyond the stated privacy exclusions.

Graph View

Backlinks