Claude Reflect: Anthropic's new usage dashboard shows you how you actually use AI

Update, 13 July 2026: Anthropic extends Fable 5 free access to July 19 and keeps Claude Code weekly limits 50% higher

Anthropic has extended included Fable 5 access on paid plans (Pro, Max, Team, and premium Enterprise seats) through July 19, 2026 at 11:59:59 PM PT, alongside a matching extension of the 50% higher weekly usage limits for Claude Code. This is the second extension in under a week: the original July 7 cutoff moved to July 12, and this announcement arrived after that deadline had already passed.

The promotion changes nothing about how limits are structured or displayed in Reflect. Fable 5 draws from the same weekly usage pool as other Claude models, so the dashboard will continue showing it as part of your normal consumption, not as a separate allocation.

Nothing needs to be activated. Free users, standard Enterprise seats, usage-based Enterprise plans, and API access remain excluded.

If the July 19 cutoff holds, continued Fable 5 use shifts to prepaid credits at $10 per million input tokens and $50 per million output tokens. Claude Code weekly limits will return to standard levels. Anthropic has stated its intention to make Fable 5 a permanent plan inclusion once compute supply allows, but has not given a timeline.

Anthropic has quietly added something genuinely useful to Claude: a personal usage dashboard called Reflect, now in beta for Free, Pro, and Max users. It lives in Settings and gives you a structured look at how you’ve been using Claude over the past one, three, six, or twelve months.

This is not a gimmick. It is a deliberate product decision, and understanding what it does and does not do is worth a few minutes of your time.

What the dashboard actually shows you

Reflect surfaces four core things: the topics you discuss most with Claude, your usage patterns over time, your most active day and peak hour, and a total chat count for your selected window. Below those headline numbers is a visual breakdown of your activity, and alongside it, observations about your working style.

That last part is where it gets more interesting. Based on your actual conversations, Reflect can surface practical suggestions. If you keep re-explaining the same project context every time you open a new chat, it might point you toward Claude’s Projects feature. If you have repetitive tasks, it may suggest building a custom workflow. It is less a statistics page and more a light coaching layer.

Reflect also periodically prompts you with a reflective question: something like “What’s one thing you want to keep doing yourself, even if Claude could do it faster?” You can answer it or dismiss it. The point is to prompt a moment of thinking about how you are using the tool, not just how often.

The 4D AI Fluency Framework

Alongside raw usage data, Reflect maps your habits against Anthropic’s 4D AI Fluency Framework. The four dimensions are:

Delegation — Are you consciously deciding when to use AI, or just defaulting to it?
Description — How effectively are you prompting Claude to get useful outputs?
Discernment — Are you critically evaluating what Claude produces?
Diligence — Are you taking responsibility for how you use the results?

This is not a simple usage counter. Someone who spends two hours a day thoughtfully working through complex problems with Claude scores differently from someone who pastes the same vague prompt twenty times and accepts whatever comes back. The framework is designed to surface qualitative patterns, not just quantity.

The wellbeing layer: quiet hours and break reminders

Inside the dashboard, under Settings > Time and Focus, you can set quiet hours during which you prefer not to use Claude, and schedule nudges to take a break after a certain amount of time. Both are optional and dismissible. Anthropic is not limiting your access; it is giving you tools to apply your own preferences.

This came directly out of user research. Ryn Linthicum, Anthropic’s head of wellbeing policy, explained that a societal impacts study found users expressing a mix of optimism and anxiety about AI products. The team built Reflect specifically to help people get more efficient at meeting their goals, not to encourage more time spent with the product. That is an unusual design brief for a consumer tech company, and it shows in the details.

What Reflect does not touch

The privacy boundaries here are explicit and worth knowing:

Incognito chats are excluded entirely. Nothing from those sessions appears in your reflection.
Health integration conversations are excluded. If you have connected a health tool, those conversations do not feed into your insights.
Source files from connected tools are excluded. If you asked Claude to summarise your inbox, the summary might appear at a high level in your reflection, but the actual emails would not.
Insights stay in Reflect. Anthropic states clearly that the information surfaced in your dashboard is not used for any other purpose.

Sensitive conversations can still appear in the reflection, but only at a high level. The system is working from patterns in your interactions, not reading back the content of individual messages.

Memory must be enabled in Settings for Reflect to work at all. If you have Memory turned off, there is nothing for the dashboard to draw from.

Current limitations and what is coming

A few things are not live yet in the beta:

Time-spent data is not currently tracked. Anthropic acknowledged this is because time-on-product was a metric they deliberately chose not to optimise for internally. They plan to add it later, but it was not available at launch.
Cowork conversations will be included in a future update.
Mobile is not yet supported. For now, Reflect is only available via the Claude web client and desktop app.

Why Anthropic built this

Anthropic’s model spec is explicit that Claude should not foster excessive engagement or reliance if that is not in the user’s genuine interest. Reflect is the product-level expression of that principle. Rather than keeping the question of “how much is too much?” entirely in the user’s head, Anthropic has built a lightweight structure to make it easier to answer.

The feature was developed with input from the MIT Media Lab’s Advancing Humans with AI programme, the Digital Wellness Lab at Boston Children’s Hospital, and the Family Online Safety Institute. That is a serious set of collaborators for what could have been a simple usage stats page.

There is also a strategic dimension. As conversations about AI dependency and data practices become louder, a publicly visible wellbeing layer is meaningful brand positioning for Anthropic. But the product decisions, including the choice not to track time spent and the explicit exclusions around health data, suggest this is more than optics.

What this means for you

If you use Claude regularly, Reflect gives you something you did not have before: a clear picture of how your usage is actually distributed. Most people have a rough sense of their habits but no real data to check that intuition against.

The topic breakdown alone can be useful. Seeing that 60% of your chats are writing tasks and 5% are the coding work you thought you were primarily using it for is the kind of concrete feedback that changes how you plan your workflow.

The 4D framework is the more ambitious part. It is asking you to think about whether you are using AI well, not just a lot. That is a harder question, and the dashboard does not claim to fully answer it. It just makes it harder to avoid asking.

To get started, go to Settings > Reflect in Claude on web or desktop. Memory needs to be on. The dashboard is in beta, so expect it to evolve.