Anthropic introduced a new beta feature for Claude called “Reflect” on July 9, 2026. It is designed to show users how they actually use the AI assistant – with which topics, at what times, and to what extent. Shortly after the official announcement from Anthropic, independent media voiced a significantly more critical assessment.
What Reflect Shows and How It Works
According to the official announcement, Reflect provides a dashboard with summaries of one’s own Claude usage: which topics dominate, at what times the AI is used, and what type of tasks prevail. Users can choose between time periods of one, three, six, or twelve months. The dashboard is complemented by periodic reflective prompts – such as the question of which task one would still prefer to do oneself, even if Claude could take over faster. Additional practical features include adjustable quiet times and reminders for usage breaks. A time tracking feature that shows how much time has been spent with Claude in total is announced by Anthropic for the future.
The 4D Fluency Framework
As a conceptual basis, Anthropic cites its own “4D AI Fluency Framework” with four competency areas: Delegation (the decision of whether and how to delegate a task to AI), Description (the ability to clearly articulate goals), Discernment (the assessment of how useful an AI result actually is), and Diligence (the responsibility for one’s own usage). Reflect is intended to categorize users based on these four dimensions and provide them with concrete improvement suggestions. According to its own statements, Anthropic developed the feature with external experts in digital well-being, including representatives from the AHA program at MIT Media Lab, the Digital Wellness Lab at Boston Children’s Hospital, and the Family Online Safety Institute.
Which Data Reflect Uses – and Which It Does Not
Reflect analyzes the chat history with Claude, according to Anthropic, but deliberately excludes certain content: Incognito chats are not included, only summaries from connected tools are considered instead of the underlying files, and conversations in the context of health integrations are excluded. Sensitive discussions are intended to be included only at the summary level in the evaluation, and the insights gained from them are supposed to remain solely within the Reflect dashboard and not be used elsewhere, according to Anthropic. The beta version is available for users of the Free, Pro, and Max plans, provided the memory function is activated, accessible through the settings in the web and desktop app. An expansion to Cowork conversations is reportedly in preparation according to Anthropic.
Criticism: Usage Analysis or Retention Tool?
While MacRumors primarily categorizes the feature descriptively, TechCrunch takes a significantly more critical stance on Reflect. Author Sarah Perez describes the feature in her headline as a tool that “quietly” convinces users of AI, and categorizes Reflect in her analysis as a retention instrument: By making visible how central Claude has already become in everyday life, switching to competing products is made more difficult. Additionally, she interprets the feature as a form of behavioral conditioning, with which Anthropic specifically trains users to integrate AI more into their workflows. As a historical comparison point, Perez references Google’s “Gmail Meter,” introduced in 2012, which similarly highlighted the centrality of Gmail in users’ digital lives – however, she believes Reflect goes a step further. This categorization is a journalistic interpretation and not a confirmed intention by Anthropic; the company itself positions Reflect solely as a tool for digital well-being.
Categorization
Reflect fits into a broader trend where AI providers increasingly want to make the usage of their products measurable and reflectable – similar to how wellness and screen time features on smartphones and social media platforms have been doing for years. Whether Reflect actually helps users to use their AI more consciously, or whether the feature primarily strengthens the bond to Claude, cannot be conclusively assessed based on the information currently available. Both interpretations do not exclude each other: A tool can simultaneously encourage more conscious usage and lower the willingness to switch. How Reflect will develop after the beta phase and whether the announced time tracking and Cowork integration will actually come remains open for now.
