Google has announced its intention to adopt Anthropic’s Model Context Protocol (MCP) to simplify connecting artificial intelligence models with enterprise data and tools. Discover how the MCP protocol enables the seamless building of applications that communicate with various information sources.
Article contents:
- Background on the MCP protocol
- Growing support from major companies
- The importance of the move for Google
- How does the MCP protocol work?
- Example of possible applications
- Challenges of widespread adoption
- What’s next?
- Frequently asked questions
Background on the MCP protocol
Anthropic developed the Model Context Protocol (MCP) to provide a standardized framework for connecting artificial intelligence models with third-party data and tools. MCP is based on the concept of “MCP servers” that expose data, and “MCP clients” that enable applications to exchange information and perform tasks bidirectionally.
Growing support from major companies
After Anthropic released MCP as an open-source project, several companies—such as Block, Apollo, Replit, Codium, and Sourcegraph—announced their adoption of it. Now, Google is joining the list, signaling increased reliance on the new standard in the world of “agentic AI,” or procedural artificial intelligence, which allows models to access data and execute commands.
The importance of the move for Google
Demis Hassabis, CEO of Google DeepMind, stated via the X platform (formerly Twitter) that the company will add MCP support to future Gemini models and its software development kit. This move can simplify connecting Google’s algorithms to enterprise data or internal systems, allowing companies to build smart solutions that go beyond simple chat to execute commands and gather information.
How does the MCP protocol work?
When an application wants to retrieve data from a specific server, the “client” requests it via the MCP protocol, receiving a structured response that enables the application to integrate it with the intelligent model. Conversely, if the interactive model needs to send instructions to move or modify specific data, it is done seamlessly through the same protocol. MCP supports bidirectional streaming, granting significant freedom in building complex workflows.
Example of possible applications
A company providing customer support via chat can use MCP to access the user database to resolve issues or execute commands such as modifying subscriptions or sending invoices. Models can also be integrated with software development systems to update code files or run tests through text commands.
Challenges of widespread adoption
Despite the advantages of this protocol, organizations must still ensure security and compatibility with their systems. Connecting AI models to sensitive data requires high security guarantees. Experts expect Anthropic, Google, and others to provide guidelines and ready-to-use solutions to facilitate the integration of MCP into various ecosystems.
What’s next?
Observers believe that the MCP protocol could become an industry standard for entering the era of agentic AI, as advanced models require actual access to data and tools. With Google joining Anthropic, the project’s momentum is growing to transform into a foundational base when building AI-driven applications, whether in the commercial or research fields.
Frequently asked questions
1. What is the MCP protocol?
It is an open-source standard that allows intelligent models to connect to various data and tools via a server-client mechanism.
2. Why is Google adopting this protocol?
To simplify connecting its models, such as Gemini, to user data and tools, enabling the creation of more interactive AI applications.
3. What are the potential benefits for companies?
They can integrate AI models into their internal systems to perform numerous tasks such as data management and technical support.
4. Are there security concerns?
Yes, connecting models to sensitive data necessitates implementing robust security policies to ensure information protection.
5. Is the protocol exclusive to Anthropic?
No, it is open-source and has already been adopted by several major companies prior to Google.
6. How will MCP change the future of smart applications?
It will allow models instant access to data and tools, paving the way toward the development of comprehensive intelligent systems.
7. What is the next step?
Additional partnerships are expected to be announced alongside the release of integrated development kits, which could make MCP the official communication standard.
