What is Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard for connecting AI applications to external systems, such as data sources, tools and workflows, so assistants can read information and take actions.
Why does Model Context Protocol (MCP) matter for marketers?
More martech vendors now expose their products as MCP servers, so an AI assistant can query or change data in a CRM or ad account. That makes permissions, logging and approval flows a marketing operations concern, not only an engineering one.
Where does Model Context Protocol (MCP) stand today?
Anthropic announced MCP on November 25, 2024. OpenAI adopted it in March 2025, and in December 2025 Anthropic donated it to the Agentic AI Foundation under the Linux Foundation. Wikipedia reports more than 10,000 MCP servers in production by mid-2026, and also records security issues, including tool poisoning, vulnerable proxy tools and an NSA report in May 2026 saying MCP's "rapid proliferation has outpaced the development of its security model".
Common use cases
- Letting an assistant read and update CRM records
- Exposing ad account actions to agents with limits
- Connecting analytics and research data to AI tools
The MCP documentation describes the protocol with an analogy: “Think of MCP like a USB-C port for AI applications.” It says AI applications like Claude or ChatGPT can use MCP to connect to data sources, tools and workflows.
Who supports it
The documentation lists assistants such as Claude and ChatGPT and development tools such as Visual Studio Code and Cursor. Wikipedia adds that Salesforce’s Headless 360 platform began routing interactions through MCP in April 2026.
In martech
ZoomInfo lists an MCP server among its products, Seismic’s developer documentation describes an MCP server in early access, and Meta has expanded its Ads MCP server, which shows vendors exposing their products this way. Wikipedia lists specification changes that now require OAuth 2.1 best practices and forbid token passthrough.