Three major AI companies—Anthropic, OpenAI, and Google—are adding Anthropic’s Model Context Protocol (MCP) to their systems. This is a major step toward creating a standard for how AI works with external tools and business software.
MCP, which Anthropic made open-source late last year, is a universal interface for AI applications. It helps models connect easily to data sources, tools, and systems without needing custom integrations for each one. The protocol has three main parts: Resources (data objects), Tools (functions for actions), and Prompts (templates that guide model behavior). To help companies adopt the protocol faster, Anthropic also provided pre-built servers for widely used enterprise software like Google Drive, GitHub, and Slack.
OpenAI announced on March 26 that it would start using MCP with its Agents SDK. CEO Sam Altman said MCP would eventually be available for ChatGPT on desktop and mobile apps, though no timeline was given.
Google followed on April 9 with an announcement from DeepMind CEO Demis Hassabis saying that Google’s Gemini models and SDK would also use MCP. Hassabis said MCP is becoming an open standard for AI agents but did not say when Google would add support.
Linking AI systems to tools and databases usually needs custom code for each connection, which takes a lot of time and is difficult to scale. MCP solves this by giving a single protocol that simplifies these connections.
Quiet a few other companies are adding MCP, including Block, Apollo, Replit, Codeium, and Sourcegraph, as reported by ZDNet.
Three major AI companies—Anthropic, OpenAI, and Google—are adding Anthropic’s Model Context Protocol (MCP) to their systems. This is a major step toward creating a standard for how AI works with external tools and business software.
MCP, which Anthropic made open-source late last year, is a universal interface for AI applications. It helps models connect easily to data sources, tools, and systems without needing custom integrations for each one. The protocol has three main parts: Resources (data objects), Tools (functions for actions), and Prompts (templates that guide model behavior). To help companies adopt the protocol faster, Anthropic also provided pre-built servers for widely used enterprise software like Google Drive, GitHub, and Slack.
OpenAI announced on March 26 that it would start using MCP with its Agents SDK. CEO Sam Altman said MCP would eventually be available for ChatGPT on desktop and mobile apps, though no timeline was given.
Google followed on April 9 with an announcement from DeepMind CEO Demis Hassabis saying that Google’s Gemini models and SDK would also use MCP. Hassabis said MCP is becoming an open standard for AI agents but did not say when Google would add support.
Linking AI systems to tools and databases usually needs custom code for each connection, which takes a lot of time and is difficult to scale. MCP solves this by giving a single protocol that simplifies these connections.
Quiet a few other companies are adding MCP, including Block, Apollo, Replit, Codeium, and Sourcegraph, as reported by ZDNet.
Source: Google, OpenAI, and Anthropic Team Up (Kind of) to Boost Business Tools AI
