Brief IA

Anthropic Revolutionizes Usage Analysis with Reflect for Claude

🛠️ AI Tools·Tom Levy·

Anthropic Revolutionizes Usage Analysis with Reflect for Claude

Anthropic Revolutionizes Usage Analysis with Reflect for Claude
Key Takeaways
1Anthropic introduces Reflect, a tool that analyzes the use of Claude over periods of up to one year.
2Reflect provides insights into the topics discussed, usage peaks, and tasks delegated to the AI.
3The feature, currently in beta, is available for Free, Pro, and Max accounts, with an extension planned for Cowork.
💡Why it mattersReflect enables users to better understand and optimize their interaction with the AI, thereby enhancing efficiency and digital accountability.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

Anthropic Unveils Reflect, a New Perspective on Using Claude

Anthropic has just launched Reflect, an innovative dashboard that allows users of Claude to closely examine their interactions with the virtual assistant. Currently in beta, Reflect provides a retrospective of the topics discussed, moments of usage, and types of tasks assigned to Claude, covering periods from one month to an entire year. Inspired by the usage retrospective concept popularized by Spotify Wrapped, Reflect invites its users to reflect on the integration of AI into their daily lives.

Detailed Analysis of Interactions with Claude

The Reflect dashboard offers an overview of user activity over periods of one, three, six, or twelve months. It highlights several key aspects:

  • The main topics discussed with Claude,
  • The times of day when the assistant is most frequently used,
  • The types of tasks commonly delegated to the AI.

Anthropic plans to soon add a feature that allows users to visualize the total time spent on Claude. Beyond these statistics, Reflect encourages critical reflection by posing questions such as: “What task do you wish to continue doing yourself, even if Claude could perform it faster?” These questions aim to engage users in a discussion with the assistant about the use of AI.

The 4D AI Fluency Framework for Optimized Collaboration

Reflect is based on an evaluation framework called “4D AI Fluency,” which examines four essential dimensions of collaboration with AI:

  • Delegation: determining when and how to use AI,
  • Description: formulating clear requests to obtain relevant answers,
  • Discernment: judging the relevance of the provided results,
  • Diligence: taking responsibility for actions performed with AI.

For each dimension, Reflect provides concrete examples, such as revising email drafts in the user's personal style or delegating tasks after defining a clear strategy. The dashboard also allows users to set quiet periods or schedule reminders to take a break after a certain duration of use, although these reminders are optional and can be ignored.

Activation and Availability of Reflect

Reflect is currently accessible in beta for users of Free, Pro, and Max accounts, provided that Claude's memory is enabled. Anthropic plans to extend this feature to users of other plans in the near future, with a specific version for Cowork also in preparation.

Users can activate Reflect through the Settings of the assistant, available on the web or desktop application. Anthropic has also announced that a version covering conversations in Claude Cowork will be available soon.

Limitations and Exclusions of Reflect

Reflect does not account for certain conversations, particularly those conducted in Incognito mode, as well as files from connected tools, such as emails from a linked inbox, or exchanges related to health tracking tools. While sensitive topics may be mentioned, they are only addressed in a general manner. Anthropic has collaborated with experts in digital well-being, including the MIT Media Lab and the Digital Wellness Lab at Boston Children's Hospital, to approach these issues with care.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.