Model Context Protocol (MCP) — an open standard for connecting AI applications to external tools and data sources, so a connection built once can be used by many different AI assistants.
What problem it solves
Before a shared standard, every connection between an AI assistant and a business system was custom: one integration for this assistant and that CRM, another for the next pairing. MCP defines a common way for a system to describe what it offers — tools it can run, data it can return — and for any compatible AI client to use it.
Where it stands
Anthropic introduced MCP in November 2024. In December 2025 it became a founding project of the Agentic AI Foundation, a fund under the Linux Foundation co-founded by Anthropic, Block and OpenAI. At that point the project reported more than 97 million monthly SDK downloads and around 10,000 active servers, with client support across ChatGPT, Claude, Gemini, Microsoft Copilot and others.
For a small business the practical meaning is simple: when a system you already use offers an MCP server, connecting it to an AI assistant stops being a custom build.
An MCP connection gives a model real access to a real system. The questions you would ask about a new employee’s access apply here too: read or write, which records, logged where. See prompt injection.
Sources: Anthropic, Introducing the Model Context Protocol, November 2024 · MCP blog, MCP joins the Agentic AI Foundation, 9 December 2025. Last checked 25 September 2026.
Related: AI agent, AI integration, Universal Commerce Protocol.