Anthropic has announced a new beta feature for Claude that lets users track their own usage patterns and reflect on whether AI is actually helping them reach their goals. The tool, called Reflect, shows up in the Settings menu on Claude's web platform and desktop app, and it's available now for Free, Pro, and Max users who have memory turned on.
The idea behind it is simple but worth paying attention to. A lot of people have started using AI tools daily without ever stopping to ask whether they're using them well, or whether they're outsourcing things they'd rather keep doing themselves. Anthropic says this came directly from user research, where one recurring theme was that people want guidance on how to actually integrate AI into their lives in a way that feels right to them.
This puts Anthropic in slightly different territory from most AI companies, which tend to focus on getting users to do more with their tools, not less. Building a feature that actively encourages people to take breaks or hold onto certain tasks themselves is an unusual move, and it signals a growing awareness in the industry that AI habit-formation is something worth designing around thoughtfully.
What the dashboard actually shows you
The Reflect dashboard gives you a breakdown of your Claude activity over the past 1, 3, 6, or 12 months. It covers:
- The key topics you've been working on
- When you tend to use Claude most during the day or week
- The types of tasks you return to most often
Soon, Anthropic says it will also add a view showing how much total time you've spent using Claude. Right now that piece isn't in the beta yet.
The tool doesn't just display data. It also asks you questions. Periodically, it will surface prompts like “What's one thing you want to keep doing yourself, even if Claude could do it faster?” and give you space to talk through your answer directly with Claude. The goal is to get users thinking critically about their relationship with AI, not just consuming usage stats.
You can set limits from inside the dashboard
Users can configure quiet hours directly in the dashboard, blocking out times when Claude won't send notifications or prompts. There's also an option to schedule a nudge after a certain amount of usage, reminding you to take a break. Both options are dismissible, so they work more like personal reminders than hard restrictions.
This kind of built-in friction is rare in consumer software. Most products are designed to maximize engagement. The fact that Anthropic is building opt-in limits into the product itself reflects a broader conversation in tech about what responsible AI design actually looks like in practice.
A framework for thinking about AI skills
The Reflect feature also ties into something Anthropic calls the 4D AI Fluency Framework. It's designed to help users think about how they work with Claude, not just how much. The four dimensions are:
- Delegation: Deciding whether and how to hand something off to AI
- Description: Clearly communicating what you need so Claude actually delivers useful results
- Discernment: Evaluating whether what Claude produces is actually good
- Diligence: Taking responsibility for what you do with AI outputs
Your reflection report gives you a summary of your Claude activity across each of these dimensions, with specific examples drawn from how you've actually used the tool. It might note, for instance, that you tend to rework Claude's email drafts in your own voice, or that you only hand off tasks after you've already worked out the strategy yourself. It also offers practical suggestions, like creating a Project to avoid re-explaining the same context repeatedly.
Privacy guardrails built into the feature
Because reflection summaries touch on real conversation history, Anthropic has put some specific privacy limits in place. A few things worth knowing:
- Incognito chats are not included in your reflection
- Underlying files from connected tools are not pulled in, only the summaries Claude generated from them
- Any conversations connected to a health integration tool are excluded entirely
- The insights in your reflection are not used for any other purpose, including training
Sensitive conversations can still appear in your reflection, but only at a high level. Anthropic says it worked with experts from MIT Media Lab's Advancing Humans with AI program, the Digital Wellness Lab at Boston Children's Hospital, and the Family Online Safety Institute while building this part of the feature.
How to get started
To generate your first report, open Settings in Claude on the web or the desktop app and select the option to reflect on your usage. If the option isn't appearing, check whether Memory is turned on in your settings because the feature requires it. Reflection for Cowork conversations is not yet available but is coming soon.
