What is MCP?

MCP is how the AI assistant your team already uses gets access to real data. Think of it as a plugin: you add it once, and from then on the assistant can look things up instead of guessing.

Why it matters

Your people already ask an AI assistant for help with plans, pitches and briefs. The assistant writes well and reasons well, but when the question is about a market, it answers from memory, and its memory is a few years of reading the internet. It sounds sure. It has never counted anything.

A plugin changes what the assistant is working with. With Hiasynth added, the same question about a market is answered from a population of 522 million people, each anchored in what Europe's statistical offices publish. The assistant still does the thinking and the writing. The numbers are now real.

For a business that means three things:

  • No new tool to learn. Nobody logs in to a dashboard or learns a query language. They ask in the chat they already have open.
  • No project to run. There is no data licence to negotiate, no integration to build, no analyst to brief. Adding the plugin takes a few minutes.
  • Answers while the meeting is still going. A market size, a shortlist of cities or an audience profile comes back in seconds, with the numbers behind it.

What it looks like at work

In a strategy meeting. Someone asks whether the Netherlands is worth entering. The assistant sizes the segment, shows where it clusters and compares it with the markets you are already in. The discussion moves on with a number instead of an opinion.

When choosing locations. You describe your best-performing sites. The assistant finds the municipalities across a country that look like them, ranks them, and tells you who lives within walking distance of each.

When briefing an agency. You describe the audience. The assistant returns the channels, subscriptions and values that set that audience apart, by region, ready to paste into the brief.

None of these needed a specialist. They needed the question.

How it works, in four lines

  • You add the plugin to your assistant once, by giving it an address and signing in.
  • You ask a question in normal words.
  • The assistant sees it has a plugin that fits, sends it a precise request, and gets numbers back.
  • It writes your answer from those numbers. You see a small "Using Hiasynth" note while it does.

MCP stands for Model Context Protocol. It is simply the standard that lets any assistant use any plugin, the way USB lets any laptop use any printer. An "MCP server" is the plugin itself, running on our machines, so there is nothing to install or maintain on yours.

  • What can it see? Only the specific request the assistant sends it. Not the conversation, not your files, not anything else the assistant knows.
  • What can it change? Nothing. All nine Hiasynth tools are read-only.
  • Is there personal data involved? No. The population is synthetic: generated from official statistics, with no real person behind any record. Results can be used, shared and published without a privacy obligation of their own.
  • Who controls access? Each person signs in with their own account and approves the connection. In Claude Team and Enterprise and in ChatGPT Business and Enterprise, an admin decides whether the plugin is available at all. Any connection can be revoked at any time.
  • Where does our data go? Nowhere. Hiasynth returns aggregates; anything you combine them with stays on your side.

Same thing, different names

Every assistant has its own word for a plugin, which is half the confusion.

Where
What they call it
Claude
Connectors
ChatGPT
Apps, found in the plugin directory
Claude Code, Cursor, most developer tools
MCP servers
Everyone else
Integrations, tools, extensions

Ready?

Getting started takes a few minutes and ends with your first answer. Pricing is one subscription. If your team wants Hiasynth inside a product or workflow of its own, Build with Hiasynth covers that, including tools we build to order.