Anthropic launched Claude Reflect in beta on July 9, 2026 — a usage dashboard built into Claude’s Settings page that shows you how, when, and why you use Claude, then asks whether any of that should change.
The product occupies an unusual space: part analytics panel, part skills coach, part digital wellness tool. For builders who spend most of their day inside Claude, there are a few details worth understanding.
What the Dashboard Shows
Reflect requires Memory to be turned on and is available to Free, Pro, and Max users on Claude’s web and desktop apps (not Team or Enterprise) — confirmed in Anthropic’s Help Center. Once enabled, it summarizes your conversation history across selectable time windows: 1, 3, 6, or 12 months.
The summary covers:
- Key topics: what you engage Claude on most often
- Task type breakdown: the kinds of work (writing, coding, analysis, brainstorming) that fill most of your sessions
- Usage patterns: when you use Claude, peak activity times, session frequency
- Coming soon: total time spent — Anthropic says “soon, we’ll add a view of how much time you’ve spent using Claude,” so this metric isn’t live yet, only announced
It’s closer to Spotify Wrapped or iOS Screen Time than a technical analytics dashboard. There’s no per-message token count, no cost breakdown, no latency data.
Privacy boundaries (per Anthropic’s announcement and Help Center article): Incognito chats are excluded, as is activity in Claude Cowork and Claude Code. Conversations tied to health integrations (e.g., Apple Health) are excluded entirely. Underlying files from connected tools don’t appear — if you asked Claude to summarize your inbox, the summary may show up but the source emails won’t. The report data stays in the dashboard and, per Anthropic, isn’t used for any other purpose (including training).
The 4D AI Fluency Framework
The part most relevant to serious Claude users: Reflect maps your usage patterns to Anthropic’s AI Fluency Framework, co-developed with academic partners at University College Cork and Ringling College, which defines skilled AI use across four dimensions.
Delegation — setting goals and deciding whether and how to involve AI at all. High delegation skill means you’re choosing the right tasks to hand off, not just reflexively reaching for Claude.
Description — prompting effectively: giving Claude the context, constraints, and output shape it needs to produce something usable. Low description skill shows up as lots of re-prompts and heavy editing.
Discernment — accurately evaluating Claude’s outputs. Knowing when to trust, when to verify, and when to override. Critical for code generation and factual work.
Diligence — taking responsibility for what gets done with Claude’s help. This includes verifying outputs before shipping, catching errors, and maintaining the human in the loop on consequential decisions.
Reflect doesn’t give you a numeric score on each dimension — it offers observations and recommendations based on detected patterns. Anthropic’s own example: if you keep re-explaining the same context for ongoing work, your reflection may suggest starting a Project instead. Engadget’s hands-on account reports Reflect also surfaced a suggestion to build a custom Skill after noticing repeated, similar questions on one topic.
Wellbeing Features
Two settings ship with the launch, per Anthropic:
- Quiet hours: block a time window during which Claude won’t respond (or will respond minimally — the exact behavior is configurable)
- Break nudges: a notification after a set amount of continuous Claude use, dismissible
Both are described by Anthropic as reminders of your own preferences, and both can be dismissed.
The dashboard also periodically surfaces a reflective prompt — for example, the exact wording Anthropic quotes: “What’s one thing you want to keep doing yourself, even if Claude could do it faster?" You can decline to engage with it, or talk it through with Claude directly.
Anthropic developed these features with researchers from 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 — named as collaborators in the launch announcement.
The Tension Worth Understanding
TechCrunch characterized Reflect as “quietly selling you on AI” — noting that mapping heavy Claude use to a “Fluency Framework” reframes what could be dependency as measurable skill. The harder-to-refute version of this critique: the recommendations Reflect surfaces (use Projects, build Skills, keep Memory on) are all actions that deepen Claude integration, not reduce it.
That’s not necessarily bad for builders, but it’s worth naming. The tool is designed to promote intentional AI use — and Anthropic’s definition of intentional use includes using Claude more effectively, not using it less.
What It Means for Builders
If you’re a heavy Claude user yourself, Reflect’s Fluency Framework gives you a vocabulary for auditing your own habits. The question “Am I using Claude where it actually helps, or where I’ve just habituated to it?” is genuinely useful to ask periodically.
If you’re building a Claude-powered product, two things are worth watching:
The upcoming Cowork conversation reflection — Anthropic’s announcement states “reflecting on your Cowork conversations will be available soon”, and as of launch, Claude Cowork and Claude Code activity is explicitly excluded from the recap. Once live, this would extend Reflect to multi-agent sessions, not just individual conversations. For builders who ship agentic workflows, that means your users may eventually get a breakdown of what their Claude-powered automations did on their behalf. Transparency features like this may become table stakes.
The recommendations model — Reflect’s observation that “you keep repeating this context, you should use Projects” is essentially a customer-success nudge embedded in the product. If you’re building SaaS on top of Claude, a similar pattern (detecting repeated friction and surfacing the right feature to solve it) could be valuable in your own product layer.
Memory requirement: Reflect only works if users have Memory turned on. If your product manages Claude’s memory externally (injecting context via system prompt rather than relying on Claude’s built-in Memory), your users won’t see their usage reflected here at all.
Access
Claude Settings → Reflect on web or desktop. Beta. Free, Pro, and Max accounts with Memory enabled — not currently available on Team or Enterprise plans.
Reflect is a consumer product, not a developer API. There’s no endpoint to query your team’s usage patterns or pull fluency scores programmatically. But for individual builders spending significant time in Claude, it’s the first tool Anthropic has shipped that turns usage history into something legible — and the Fluency Framework is a useful mental model regardless of whether you read your own Reflect report.