<aside> 🔌

Module 2 — Connecting to Your World

So far, Claude lives on your computer. MCPs are what connect it to everything else. This page explains what they are and why they matter.

</aside>

<aside> 📍

Page 5 of 14 · Module 2 — Connecting to Your World

</aside>

The Problem With AI Assistants

Every AI assistant has the same limitation: it can only work with what you give it.

You want help with an email? You have to copy the email in.

You want to update your Notion? You have to copy the content out, get Claude to rewrite it, then paste it back.

You want to check your calendar? You have to tell Claude what's in it.

It's powerful, but it's manual. You're still doing the connecting.

MCPs fix this.

What Is an MCP?

MCP stands for Model Context Protocol — but ignore the jargon. Here's what it actually means:

An MCP is a connection between Claude and one of your tools.

When you connect an MCP, Claude gets direct access to that tool. Not read-only access to a summary — actual access. It can read, write, search, create, and update, just like you can.

flowchart LR
    subgraph Before ["❌ Without MCPs"]
        U1["You"] -->|"copy text manually"| C1["Claude"]
        C1 -->|"gives answer"| U1
        U1 -->|"paste back manually"| T1["Notion / Gmail / Slack"]
    end
    subgraph After ["✅ With MCPs"]
        U2["You"] -->|"just ask"| C2["Claude"]
        C2 <-->|"reads & writes directly"| N["Notion"]
        C2 <-->|"reads & writes directly"| G["Gmail"]
        C2 <-->|"reads & writes directly"| S["Slack"]
        C2 -->|"result"| U2
    end

What This Looks Like in Practice

Here's the before and after: