What is MCP? The Model Context Protocol explained for your business

MCP, short for Model Context Protocol, is an open standard for connecting AI applications to other systems: files, databases, your CRM or your CMS. With MCP, an AI agent can look up data and carry out tasks in software you already use. Nobody has to build a separate integration for every AI tool.

So what is MCP in practice for your business? This article explains how MCP works, what an MCP server is and how MCP differs from a regular API. You will also read what it means for businesses in Belgium and the Netherlands that want to connect an agent to their software. And you will see where the risks are and how to keep permissions tight.

A dark gray central block with four thin blue lines running to four different stone and graphite shapes
One standard, and a separate connection with its own permissions for each system.

Where does MCP come from?

Anthropic, the company behind Claude, introduced MCP in November 2024 as an open standard. In December 2025, Anthropic donated the protocol to the Agentic AI Foundation, a fund under the Linux Foundation. That way, no single AI vendor controls it. According to that announcement, there were more than 10,000 active public MCP servers at the time. ChatGPT, Cursor, Gemini, Microsoft Copilot and Visual Studio Code were among the products that supported MCP.

The official documentation puts it this way: “Think of MCP like a USB-C port for AI applications.” One kind of connector that many devices can plug into. With MCP, those devices are your systems.

How does MCP work, and what is an MCP server?

MCP has three roles. The host is the AI application you work in, for example an AI agent like Hermes. For every system the host connects to, it starts an MCP client. On the other side is the MCP server: a program that makes one system available to AI. So you have one server per system, for example one for your CRM and one for your documents.

According to the architecture overview, an MCP server offers three kinds of things. Tools are actions the AI can take, such as looking up a customer or creating a task. Resources are data that serve as context, such as a file or the structure of a database. Prompts are reusable templates for recurring tasks.

At startup, the client asks each server which tools it has. The model sees that list and picks the tool that fits your request. The client runs the tool and returns the result. An MCP server can run locally, on the same computer as the agent, or remotely. Some software vendors offer such a remote server themselves.

What is the difference between MCP and an API?

An API is the entry point of one software package: every package has its own, with its own rules. MCP is a shared language between AI applications and those systems. Behind the scenes, an MCP server often uses the software’s own API. The difference is who has to understand the integration. Without MCP, someone writes code for every API in every AI application. With MCP, you build a server once, and every application that supports MCP can use it.

FeatureAPIMCP

What is it?

The entry point of one software package

An open standard for connecting AI applications to systems

Built for

Programs and developers

AI applications and AI agents

How does the AI find out what it can do?

Someone reads the documentation and writes code

The client asks the server which tools are available

Reuse

A separate integration for each application

One server works with any MCP client

Permissions

Through the key or account of the API

Through the server’s key, plus the tools you allow

So MCP does not replace an API. If your CRM has a good API but no MCP server, an agent can use that API directly. Hermes does that when a task calls for it.

What does MCP do for your business?

The biggest benefit is reuse. If an MCP server already exists for your CRM, your project software or your database, nobody has to build the integration from scratch. If you later switch to a different AI application, that server still works.

You also get clearer boundaries. Each MCP server is a separate connection with its own key. You decide per system what the agent may do, and you can switch off one connection without touching the rest.

And an agent can carry out one task across several systems. For a quote request, it looks up the customer in your CRM, reads earlier agreements in your documents and creates a task in your project software. Which integrations make sense and which ones to skip is covered in connecting AI to your CRM and CMS. How we set up these integrations for businesses is described on the AI automation page. You capture how the agent handles such a task, step by step, in a skill. Read more in agent skills.

How does Hermes use MCP?

Hermes, the open-source AI agent Voltti works with, supports MCP out of the box. You add an MCP server to the configuration. At startup, Hermes discovers that server’s tools and registers them. The Hermes documentation on MCP explains how you choose, per server, which tools the agent gets to see. You either list the tools you allow, or list the tools you switch off. For a payment system, that means you can allow looking up payments but not issuing refunds. A local MCP server also does not get all your environment variables by default, only what you configure plus a safe baseline.

Our own Hermes installation at Voltti works this way with Storyblok, the CMS behind this website, and with Search Console. Each system has its own access. Search Console is read-only: the agent pulls search data from it but cannot change anything there. In Storyblok, the agent edits content and in some cases publishes it too. We review the work afterward and step in where needed. To see what that looks like in practice, read retrieving marketing data with an AI agent and SEO with an AI agent.

What are the risks of MCP?

MCP is a protocol. How safe an integration is depends on the server and on the permissions you grant. Pay attention to these four points:

  • Permissions that are too broad. A server with a key that can do everything gives the agent everything. Create a separate key for each server with only the permissions the task needs.

  • Hidden instructions. A document, email or web page the agent reads through a server can contain instructions. That is called prompt injection. Read more in prompt injection and API keys.

  • Third-party code. An MCP server is software that runs on your systems. Only install servers whose source you know, and check what a server runs during installation.

  • Too many tools. Every tool the agent sees is a possible action. Switch off tools you do not need, and have a person approve actions with major consequences first.

The official documentation has a separate page with security best practices for MCP, mainly for people who build servers. For a practical list for your own business, see our safety checklist for AI and business data. How we protect data and access is explained on the security and privacy page.

What if your software has no MCP server?

Then there are other routes. If the software has an API, the agent can use it. If there is no API and no MCP server, we build a custom integration ourselves. The form depends on what the system allows. It can be a small connector with its own API endpoint. The agent can also read exports, such as a CSV file or a scheduled report. Or the agent operates the web interface itself, through browser automation. Even then, the agent only gets the permissions the task needs.

Where do you start?

Start with the work. Pick one task that comes back every week and look at which systems the agent needs for it. Only then do you choose the route: an existing MCP server, the API or a custom integration. In a pilot, you test that one integration alongside your current way of working. A pilot usually takes 2 to 4 weeks, from the first conversation to the evaluation. How that works is described in our approach.

Portrait of Seppe Gadeyne

Seppe Gadeyne is a developer and the owner of Voltti. He sets up AI agents for companies in Belgium and the Netherlands.

Updated on

Frequently asked questions

01Is MCP the same as an API?

No. An API is the entry point of one software package, with its own rules. MCP is an open standard that lets AI applications talk to many different systems in the same way. An MCP server often uses the API of the software it makes available.

02Who created MCP?

Anthropic, the company behind Claude, introduced MCP in November 2024. Since December 2025, the protocol has been part of the Agentic AI Foundation, a fund under the Linux Foundation. MCP is open source and not tied to a single AI vendor.

03What is an MCP server?

A program that makes one system available to AI applications, for example your CRM, a database or your files. The server offers tools (actions), resources (data) and prompts (templates). An MCP server can run locally or remotely, for example at the software vendor.

04Is an MCP server safe?

That depends on the server and on the permissions you give it. Use servers whose source you know and give each server its own key with as few permissions as possible. Switch off tools the agent does not need. Have a person approve actions with major consequences first.

05Do you need MCP to use an AI agent?

No. MCP is one way to connect an agent to your software. An agent can also use an API directly, read exports or work through a custom integration. MCP is most useful when a server already exists for the software you use.

Which software do you want to connect to an AI agent?

Discuss your pilot