Glossary · MCP
What is MCP (Model Context Protocol)?
By Kristian Stig Henriksen · Dear Future
MCP (Model Context Protocol) is an open standard that lets AI assistants connect safely to external systems: your data, your tools, your numbers. It was introduced by Anthropic in November 2024 and has since been adopted across the major assistants. The shortest explanation is an analogy: MCP is USB-C for AI. One shared plug, instead of a new custom integration for every pair of assistant and system.
What does MCP mean?
The name says it all. Model: the language model behind your assistant. Context: what the model gets access to beyond its training, meaning your systems and data. Protocol: a shared language for that access. Before MCP, every tool had to be integrated separately with every assistant; with MCP, a system exposes one server, and any assistant that speaks the standard can connect. It is exactly what USB-C did to the cable drawer.
How MCP works in practice
A system exposes an MCP server with a set of named tools: fetch numbers, search data, create a task. Your assistant sees the tools and can use them when you ask, with your access and your permissions. The answer is then built on an actual lookup in the system, not on the model's memory. It is the difference between asking someone who guesses well and someone who just looked it up.
Why MCP makes a measurement tool stronger
A measurement is only worth the decisions it moves, and decisions are made where you work. Our AI Visibility Engine therefore ships with its own MCP server: you can ask about your own visibility numbers in plain language, inside your own AI assistant, instead of logging into yet another dashboard.
- Ask freely: “where are we losing citations to competitors?” gets answered with your real numbers, fetched that second.
- Act immediately: log opportunities and tasks into your roadmap from the conversation, so insight becomes plan without switching tools.
- Safely scoped: access is per user and limited to your workspace, so the assistant only sees what you yourself can access.
The combination is the point: the daily measurement makes the numbers trustworthy, and MCP makes them available exactly when the question arises. It closes the distance between knowing and doing.
Also read what an LLM is and the guide to measuring AI visibility, or see the method behind the numbers.
Try it on your own numbers
Your visibility, available in your own assistant
Get an AI visibility analysis with MCP access included: daily measured numbers you can query in plain language from day one.
Get the analysis →Frequently asked questions
What does MCP stand for?
MCP stands for Model Context Protocol: an open standard that gives AI assistants a shared, safe way to connect to external systems and data. The analogy is USB-C: one shared plug instead of a custom integration for every combination of assistant and tool.
Is MCP safe?
The standard is built on explicit access: you choose which servers your assistant may use, and the server decides what each user may see and do. In our MCP server, access is per user and scoped to your workspace. As with all access control, safety depends on the implementation, so ask any vendor how their server scopes access.
Which AI assistants support MCP?
The standard is open and broadly adopted: it is used by the major assistants and by a growing set of developer tools and business systems. In practice that means the same MCP lookup can be used from several assistants, with no new integration for each.