Claude's New Reflect Feature Shows You What You Actually Use AI For
Anthropic's new Reflect dashboard tracks how you use Claude — your topics, patterns, and habits — and asks a rare question for a tech product: is this how you want to spend your time?
Anthropic has launched Reflect, a new beta feature that turns the mirror around: instead of Claude analyzing your documents, it analyzes how you use Claude. The dashboard, found in Settings on the web or desktop app, summarizes what you’ve been working on, which topics come up most, and when you tend to reach for AI — looking back over the past 1, 3, 6, or 12 months. A view showing how much time you’ve spent with Claude is coming soon.
Think of it as Spotify Wrapped for your AI use, but with a twist: it occasionally asks whether you’re happy with what it finds. Reflect 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 that through with Claude. You can also set quiet hours or schedule a nudge to take a break after a certain amount of use. The feature builds on what Anthropic calls the 4D AI Fluency Framework: delegation (deciding what to hand over), description (asking well), discernment (judging the output), and diligence (owning the result). Your report shows how you tend to collaborate across those four — noting, say, that you rework email drafts in your own voice, or only delegate once you’ve settled the strategy yourself.
On privacy: reflections skip incognito chats entirely, don’t pull in files from connected tools (a summary of your inbox may appear, the actual emails won’t), and anything tied to a health integration stays out completely. Anthropic says the insights aren’t used for any other purpose, and that it worked with wellbeing researchers from the MIT Media Lab, Boston Children’s Hospital’s Digital Wellness Lab, and the Family Online Safety Institute on the design.
What’s behind this? Partly a real answer to a real question — Anthropic’s user interviews kept surfacing the same uncertainty about how much AI use is the right amount. But TechCrunch makes a fair point: a dashboard showing everything AI does for you is also quietly a sales pitch, the way a fitness tracker makes you want to close your rings. A tool that says “you used Claude for 40 tasks this month” cuts both ways.
What this means for you: If you use Claude regularly, Reflect is worth ten minutes — most of us have no accurate picture of what we’ve actually outsourced, and seeing it laid out can be genuinely clarifying. It’s free on all plans, including the free tier, but only works if you have Memory turned on — which is a trade-off in itself: you’re sharing more history with Claude to see how you use Claude. Decide whether that mirror is worth it to you.
Sources
One Million Lines in 11 Days: AI Rewrote the Entire Bun Runtime in Rust
The JavaScript tool Bun has been fully rewritten from Zig to Rust — and Claude Fable 5 did most of the work, running 64 instances in parallel for 11 days at a cost of $165,000.
