MCP

Model Context Protocol

In one sentenceAn open standard that lets AI assistants plug into your tools and data (CRM, calendar, files, databases) through one common connector.

What it means

Before MCP, every AI app needed a custom integration for every tool. MCP is a shared standard, often described as "USB-C for AI": a tool builds one MCP server, and any AI app that supports MCP can use it.

Anthropic introduced MCP in November 2024 and it has since been adopted across the industry, including by OpenAI, Google and Microsoft. In Claude, ChatGPT and other apps you will often see MCP servers called "connectors".

MCP vs API: an API is how software talks to software, and every app has its own. MCP sits on top of APIs and gives AI a standard way to discover and use them. An MCP server usually calls the tool's API behind the scenes.

How to use it

• Look for "connectors" or "integrations" in your AI app's settings and connect the tools you use every day: Gmail, Google Calendar, Google Drive, Notion, HubSpot, Slack.
• Then ask questions across them: "Who did I meet last week that I have not followed up with?"
• Only connect tools from sources you trust, and check what permissions each connector asks for. A connector can read (and sometimes change) your data.

Look at my calendar for last week and my sent email. List everyone I met with but have not emailed since, with a one-line suggested follow-up for each.

Related terms

API AI Agent Tool Use Integration Prompt Injection