Model Context Protocol (MCP)
Definition
Model Context Protocol (MCP) is an open standard that lets AI apps connect to outside tools and data, such as a calendar, a CRM or a database, through one shared format. Apps that support MCP can then use those tools without a custom integration for each one.
By Ronan Liedmeier, fractional head of growth and course author · Published 17 October 2026 · Last updated 17 October 2026
Also called: MCP, MCP connectors
Anthropic introduced it in November 2024, and the official site compares it to a USB-C port for AI applications. At work, MCP is how an assistant goes from answering questions to reading your inbox or updating a Notion page.
How it works
There are two sides. An MCP server exposes a service's actions and data, for example "search emails" or "create a contact". An MCP client, which is the AI app such as Claude or ChatGPT, connects to that server and hands those actions to the model as tools. Each service needs only one server, and clients that support MCP can connect to it.
For most people, connecting a server is a sign-in and an approval screen.
Example
A marketing lead connects her CRM's MCP server and Google Calendar to her AI assistant. Before each sales call, she asks for the contact's open deals and recent support tickets next to the meeting agenda. The CRM server can also edit records, so she lets the assistant read freely and approves each change it proposes.
Common mistake
Connecting every account you have and letting the AI send, delete or update without a check. Connect only what a task needs, keep writes behind approval, and remember that an email or document the AI reads can contain hidden instructions meant to hijack it, a risk known as prompt injection.
Learn it in AI Academy
- MCP: Plugging AI Into Your Actual ToolsBeginner · 12 min
- MCP: The Integration LayerExpert · 8 min
Related terms
- MCP server
- An MCP server is a small program that connects an AI assistant to one tool or data source, such as Notion, Gmail or a database, using the Model Context Protocol. It tells the assistant what it can read or do there and carries out those actions when asked.
- Tool use (function calling)
- Tool use, also called function calling, is when an AI model asks your software to run a specific action, such as a search or a database lookup, and then uses the result in its answer. The model chooses the tool and the inputs; your code runs it.
- MCP gateway
- An MCP gateway is a single entry point that sits between your AI agents and all of your MCP servers. It holds the passwords and keys, decides which tools each agent may use, and keeps one log of every call.
- Prompt injection
- Prompt injection is an attack where text an AI reads, such as an email, web page or file, contains instructions that make the model do something its user did not intend. OWASP ranks it first in its Top 10 risks for LLM applications.