Jargon Buster / Model Context Protocol (MCP)
Model Context Protocol (MCP).
AI & AutomationAn open standard that lets AI assistants connect to business tools and data through one common method.
What is the Model Context Protocol?
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to outside tools and data sources, such as a CRM, a file store or a database, through one common method.
Anthropic introduced MCP in November 2024. In December 2025 it was donated to the Agentic AI Foundation under the Linux Foundation, and it is supported by major AI platforms including ChatGPT, Claude, Gemini and Microsoft Copilot.
Why it matters
Before MCP, connecting an AI assistant to each of your systems usually meant a bespoke build for every pairing. With a shared standard, a software supplier can publish one MCP server and any compatible assistant can use it. For a business, that means your AI assistant can look up a customer, read a project folder or draft an update in your systems without custom development, and you are less tied to a single AI vendor.
How it works
- The AI app The assistant you use, such as a desktop AI app, acts as the client that connects out to other systems.
- MCP servers Small connectors, often published by the software supplier, that expose a system to the AI in a standard way.
- Tools for actions Actions the server offers, such as 'search contacts' or 'create task', which the AI can choose to call.
- Resources for context Data the server can share, such as files or records, for the AI to read as context.
- Your permissions You approve the connection and control what the AI may read and do.
The honest view
MCP is plumbing, and good plumbing. Its weakness is the same as its strength: it makes connecting AI to your systems so easy that it is equally easy to give an assistant far more access than the task needs.
Third-party MCP servers vary widely in quality, and a connected assistant can be tricked by malicious instructions hidden in an email or web page it reads. Treat each connection as you would a new member of staff with system access: official sources only, the minimum permissions, and write access granted only where it is genuinely needed.
What to do about it
List your core systems and check which ones publish an official MCP server or connector. Start with read-only access to one system, such as your file store or CRM, and see what your team actually uses it for. Avoid unofficial servers from unknown developers. Keep a simple register of which AI tools are connected to what, with which permissions, and review it whenever staff or suppliers change.
Related terms.
AI Agent
Software that uses an AI model to work towards a goal, choosing its own steps and using tools to get there.
AI & AutomationSee more AAPI
A set of rules that lets one piece of software ask another for data or to carry out an action.
AI & AutomationSee more IIntegration
A connection between software systems that lets them share data or trigger actions without manual copying.
AI & AutomationSee more LLarge Language Model (LLM)
An AI system trained on huge amounts of text that can write, summarise, answer questions and follow instructions.
AI & AutomationSee moreGot a question about your own setup? Let's talk.
Your first call is with our founder, Rhodri Williams.