Also known as: MCP
In plain English
MCP is like a universal adapter for AI. Instead of building a custom connection for every app, a tool built once with MCP can plug into many different AI assistants.
In practice
MCP lets you connect assistants to systems such as file stores, ticketing tools, CRMs and databases through ready-made or custom "MCP servers". Treat each server as third-party software: review what it can access, and vet community-built servers before connecting them to sensitive systems.
Under the hood
Introduced by Anthropic in November 2024 and since adopted widely across the industry, MCP defines a client-server protocol through which servers expose tools, resources and prompts to AI applications. It standardises discovery and invocation, so integrations become reusable across different AI hosts.
Example
"We built an MCP server so the assistant can read our Confluence space."