OpenAI Agents Introduce Support for Anthropic’s Protocol
OpenAI Integrates Anthropic’s Model Context Protocol
OpenAI has recently announced its support for Anthropic’s Model Context Protocol (MCP), a new open-source standard designed to connect artificial intelligence (AI) applications to various data repositories. This integration is a significant step for developers, allowing them to ensure that their chatbots provide responses grounded in relevant information.
On a recent Wednesday, OpenAI CEO Sam Altman shared this development via a post on X. He highlighted that MCP can now be utilized within the Agents SDK, a toolkit for developers crafting AI-driven agents. The OpenAI developer team is currently working on enhancing support for the ChatGPT desktop application and the OpenAI API, with more updates expected in the near future.
Understanding the Functionality of Model Context Protocol (MCP)
AI model scaling often presents challenges, primarily due to the necessity of accessing new third-party data sources. These sources, whether they are cloud servers, business platforms, or legacy systems, have different technological frameworks which typically require developers to create custom integrations for each system.
The Role of MCP
MCP serves as a universal language, facilitating the creation of two-way connections between various AI models and third-party data sources. According to Anthropic, this leads to a simpler and more reliable method for AI systems to access necessary data.
With the Agents SDK, developers can either create “MCP servers” to expose their data or build “MCP clients,” applications designed to connect seamlessly to these servers. This eliminates the need for numerous custom integrations while providing AI with the context it needs across different tools.
Mike Krieger, Chief Product Officer at Anthropic, expressed his enthusiasm regarding the integration of MCP with OpenAI, stating in a post on X that large language models (LLMs) are most effective when they can connect to existing data and software resources.
Current AI Platforms Supporting MCP
MCP was made open-source in November, and since then, several AI platforms like Microsoft Copilot, Apollo, Replit, Codeium, and Sourcegraph have embraced the protocol.
Microsoft’s Adoption of MCP
Just recently, Microsoft announced MCP support in Copilot Studio, permitting users to connect their agents to MCP servers directly within the platform. Following OpenAI’s announcement, Microsoft also introduced a Playwright MCP server on GitHub. This server allows MCP agents to quickly access automated browser capabilities, such as clicking and typing, which enhances their functionality.
The Implications of Widespread MCP Adoption
The increasing number of AI companies that adopt MCP enhances its value for the AI community as a whole. With OpenAI—a significant competitor to Anthropic—now onboard, there is potential for Anthropic to establish itself as the leading force in creating a global standard for interconnectivity among AI agents. This progression could encourage further collaboration and compatibility across diverse AI platforms.
