Introduction to MCP

You have probably used an AI assistant before.

You type a question, and the AI gives you an answer.

For example:

What is 25 × 40?

The AI can answer:

1000

Pretty simple.

But now imagine that the AI needs to use a calculator to get the answer.

The calculator is a separate program. The AI needs some way to communicate with it.

So we now have two things:

This looks simple.

But now think about something bigger.

What if we want our AI assistant to use many different things?

For example:

How should the AI communicate with all of them?

This is the problem that MCP helps solve.

💡 What is MCP?

MCP stands for Model Context Protocol.

MCP is a standard way for an AI application to communicate with external tools and data.

That may sound complicated, so let’s make it simpler.

Think of MCP as a common language that an AI application and external systems can use to communicate.

Instead of creating a completely different communication method for every tool, MCP provides a common way to do it.

In simple words:

MCP gives AI a standard way to use external tools and access external information.

That’s the main idea.

We will understand exactly how it works as we build our project.

🧩 What Problem Does MCP Solve?

Let’s forget MCP for a moment.

Imagine we have an AI assistant that needs to use three different things:

Without a common standard, the AI application may need a different integration for each one.

Conceptually, it could look like this:

Now imagine adding 20 more systems.

The application becomes harder to build and maintain.

And another AI application may need to build the same integrations again.

We need a better approach.

🔌 Why Do We Need a Standard?

Let’s take a simple example from everyday life.

Imagine that every mobile phone had a completely different charging connector.

If you buy a new phone, you may need to buy a completely different charger.

A common standard makes this easier.

The same basic idea applies to MCP.

Instead of every AI application and every external system inventing its own way of communicating, MCP provides a standard way of communicating.

Conceptually:

With MCP:

The actual MCP architecture is more detailed than this. We will build it step by step and understand every part.

🗣️ MCP in Simple Words

If you remember only one thing from this page, remember this:

MCP is a standard way for AI applications to connect to and use external capabilities.

For example, an AI application could use MCP to:

• Perform a calculation
• Read a file
• Search information
• Get data from another system
• Perform an operation
• Access information from an external source

The important point is not the calculator or the file.

The important point is that MCP defines a common way for the AI application to communicate with these capabilities.

🌍 A Simple Real-World Example

Let’s imagine we have an AI assistant called My Assistant.

You ask:

“Calculate 125 × 48.”

The AI understands your request.

It decides that it needs a calculator.

A simplified flow could look like this:

The AI can then tell you:

125 × 48 is 6000.

Later, our calculator will not be the only thing available.

We might have:

And our AI application can communicate with these capabilities through MCP.

🖥️ What Is an MCP Server?

You will hear the term MCP Server throughout this tutorial.

For now, don’t worry about the technical details.

Think of an MCP server as a program that provides capabilities to an AI application through MCP.

For example, we could create an MCP server that provides a calculator.

The MCP server can tell the AI application:

“I have a calculator capability.”

The AI application can then ask:

“Please use the calculator to multiply 25 and 40.”

We will build this ourselves.

And once we build it, the idea will become much clearer.

🛠️ What Are We Going to Build?

Rather than learning MCP only through theory, we are going to learn it by building things.

We will start with a very small project.

Our first goal is:

We will create an MCP server and expose a simple calculator operation through it.

For example:

add(10, 20)

Then we will gradually make our project more capable.

So we are not going to learn MCP as a collection of definitions.

We are going to build it.

✅ One Last Thing

You don’t need to understand everything about MCP before continuing.

At this point, you only need to remember:

That’s enough for now.

Before we write code, we need to understand the basic MCP architecture.

In the next chapter, we will learn who actually talks to whom in MCP.