Anthropic Launches MCP: A New Era For AI Connected To Applications

Anthropic unveils the Model Context Protocol (MCP), an open-source standard that revolutionizes the connection between AI and business applications. This standardized protocol facilitates model access like Claude to external data, paving the way for truly integrated intelligent assistants within workflows.

A persistent problem: the isolation of AIs

Language models like Claude, ChatGPT, or Gemini excel in conversation, but suffer from a major problem: they remain isolated from business tools. Whether for code management, database querying, or interaction with tools like Slack, integrations remain specific and complex to maintain. This fragmentation limits the capabilities of AIs and hinders their adoption in the professional world.

With MCP, Anthropic offers a universal solution that simplifies the connection between AIs and existing applications.

MCP: An open protocol to connect Claude to applications

MCP is an open protocol that allows developers to create standard connectors between AI models and various digital services. Specifically, MCP relies on a client-server architecture:

• MCP Servers: they expose data sources or specific features (e.g. access to Google Drive files, Slack message management, querying Postgres databases).
• MCP Clients: AI applications like Claude Desktop can connect to it to retrieve context and perform actions.

The goal is to simplify the integration of AIs with professional tools without requiring complex custom integrations.

Fleur MCP: an interface to democratize connections

If MCP remains a powerful technology, its technical implementation can deter non-developers. This is where Fleur MCP comes in, an application developed by an independent team.

Fleur works like an App Store for MCP connectors: instead of manually coding integrations, users can install connectors with one click. Some examples available:

• GitHub: querying repositories, creating tickets, commenting on code.
• Browser: automated web browsing, form filling.
• Slack: retrieving and summarizing conversations.
• Obsidian: searching and managing personal notes.

Fleur is currently only available on macOS, but its potential could accelerate the adoption of MCP.

Concrete and promising use cases

Companies like Block and Apollo have already adopted MCP to connect their internal systems to Claude. Among the most promising use cases:

• Software development: code generation and analysis with GitHub.
• Task automation: direct interaction with project management tools like Linear.
• Information search: automated web browsing and retrieval of relevant data.

Thanks to MCP, AI models are no longer just dialogue boxes. They can act, interact with applications, and automate workflows.

Towards a more autonomous and integrated AI

With MCP, Anthropic adopts a modular approach, similar to Unix: instead of developing an AI that can do everything, the goal is to connect the AI to the right tools.

We can imagine that in the future, Claude could:

• Book an Uber.
• Manage a professional calendar.
• Supervise a complete project on GitHub and Slack.

This evolution could mark a major turning point in the use of conversational AI. Thanks to MCP and Fleur, Claude becomes more than an assistant: it transforms into a true digital operator.