Brief IA

Claude from Anthropic: The AI Toolkit Challenging ChatGPT

🛠️ AI Tools·Tom Levy·

Claude from Anthropic: The AI Toolkit Challenging ChatGPT

Claude from Anthropic: The AI Toolkit Challenging ChatGPT
Key Takeaways
1Launched in France two years ago, Claude by Anthropic is establishing itself against ChatGPT with a variety of tools.
2Claude Code and Cowork, two flagship solutions, generate billions of dollars in revenue for Anthropic.
3The Claude models, such as Haiku and Fable, adapt to tasks ranging from simple classification to advanced cybersecurity.
💡Why it mattersAnthropic is diversifying its offerings to meet the varied needs of businesses, solidifying its position in the AI market.
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

The Different Tools Offered by Anthropic

Since its launch in France just over two years ago, Claude has quickly established itself as a serious competitor to ChatGPT, especially in the professional domain. Anthropic, the company behind Claude, has since expanded its offerings with a multitude of products, models, and features to meet the diverse needs of its users.

The Autonomous Tools Offered by Anthropic

Anthropic offers several autonomous solutions that stand out for their diversity and effectiveness:

  • Claude, the conversational assistant, is accessible via a browser, iOS and Android apps, as well as desktop applications for Mac and Windows. This tool allows users to formulate queries to which Claude responds by analyzing documents, drafting text, or producing code.

  • Claude Code is a tool dedicated to AI-assisted coding. It operates in the terminal, in IDEs like VS Code and JetBrains, as well as in the desktop application or via Slack. It can analyze an entire codebase, coordinate changes across multiple files, and integrate with Git, GitHub, and GitLab to manage workflow. Although it was launched quietly, Claude Code has become one of Anthropic's most popular products, generating several billion dollars in annual revenue.

  • Claude Cowork is designed for office tasks and is based on the same technology as Claude Code, but targets a non-technical audience. Users can grant access to a folder, describe an expected outcome in natural language, and Claude executes the task, whether it involves sorting files, producing a report from scattered data, or generating a presentation. Initially limited to the desktop application, Cowork is now available on the web and mobile.

Claude Integrations with Existing Tools

In addition to its native solutions, Anthropic integrates Claude into widely used software and services:

  • Claude for Chrome allows direct interaction with web pages through a beta extension for paying users. This solution enables Claude to navigate, click, and fill out forms on the web.

  • Claude for Microsoft 365 offers add-ins for Excel, PowerPoint, and Word, with a beta version for Outlook. These integrations share a unique conversation thread, allowing users to start an analysis in Excel and integrate it directly into PowerPoint presentations or Word documents without having to re-explain the context at each step.

  • Claude Tag on Slack provides the ability to delegate tasks to Claude by mentioning it in a conversation, thus facilitating team collaboration.

  • Claude Design is a visual creation tool that generates mockups, interactive prototypes, presentations, or marketing materials from a prompt. It allows for export to Canva, PDF, PPTX, or to Claude Code for implementation. Introduced in April 2026 by Anthropic Labs, it is accessible to Pro, Max, Team, and Enterprise subscribers.

Which Models to Use Based on Tasks to Be Performed?

Anthropic also offers a range of models tailored to the complexity of tasks to be performed:

  • Claude Haiku is the fastest model, ideal for simple and bulk tasks such as classification, extraction, or short synthesis. The current most performant version is Haiku 4.5.

  • Claude Sonnet provides a good balance between speed and intelligence. It became the default model at the end of June 2026 and is recommended for the majority of uses. The current most performant version is Sonnet 5.

  • Claude Opus is designed for complex tasks, including agentic coding and enterprise work. It features an effort tuning that adjusts the model's depth of thought and token consumption. The current most performant version is Opus 5.

  • Claude Fable is the most powerful model, introduced in June 2026. It is described by Anthropic as being designed for the most demanding reasoning tasks and long-duration agentic tasks. The current most performant version is Fable 5.

  • Claude Mythos, while powerful, is a variant of Fable focused on defensive cybersecurity. Its access is extremely limited and reserved for a few companies participating in the Project Glasswing program.

Models from previous generations, such as Opus 4.7 or Sonnet 4.6, remain available but are now considered "legacy" models, meaning they are no longer the most recent recommendations.

Understanding the Features Offered by Anthropic in Claude

In addition to tools and models, Claude integrates a series of advanced features that extend, connect, and customize its capabilities:

  • Skills are instruction folders that transform Claude into a specialist on a specific task. They load on demand and work within the application, in Claude Code, or via the API.

  • Artifacts are interactive outputs generated in conversation, such as dashboards, mini-apps, or visualizations. They support connectors, persistent storage, and API calls, and can be easily shared via a dedicated URL.

  • Projects are dedicated workspaces for a subject or recurring mission. They group reference files, instructions, and context specific to that work, and Claude remembers them from one conversation to the next within the project. Each project has its own memory, distinct from Claude's global memory and other projects.

  • Connectors (MCP), based on the open protocol Model Context Protocol, link Claude to external services like Slack, Google Drive, or Gmail. Anthropic has recently expanded this directory to creative tools such as Adobe, Canva, and Blender.

  • Memory is grouped in the settings under the Capabilities tab and defines what Claude retains between conversations, with the ability to review, correct, or delete this information. Anthropic has also added Reflect, a dashboard that summarizes discussed topics and delegated tasks.

  • Computer use allows Cowork and Claude Code to directly control the computer, using the keyboard and mouse. This capability remains in research preview and requires particular vigilance, with Anthropic warning against the risks of prompt injection.

This ecosystem is evolving rapidly, with new tools and models announced regularly.

Brief IA — L'actualité IA en français

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