Learn by clicking · ~3 min read · Day 22 of 30

What is an "API," and why should you care?

An API (Application Programming Interface) is how one piece of software talks to another. Think of it as a waiter: you do not walk into the kitchen and cook, you give the waiter a clear order, and they bring back exactly what you asked for. An API is that waiter for software.

An app (you)

a website, a phone app, your AI assistant…

→sends a
request

The API

a menu of requests it will answer

getWeather()getPrice()sendEmail()

→returns a
response

The kitchen

the database, service, or system behind it

📋 A defined menu of allowed requests 🤝 App to app: software talking to software 📦 Answers come back as tidy data 🔁 Reusable: same request, same shape, every time

Every time your phone shows the weather, a map loads, or you pay online, an API is carrying that order back and forth. The menu matters: an API only answers the requests it was built to answer. You will send a real one yourself in a minute.

02 / 05 · The myth, busted

"That is just programmer stuff. And isn't it the same as an MCP?"

Two myths, and both are worth killing early.

Myth 1: "APIs are only for programmers, irrelevant to me." Not quite. You do not have to write code to care about APIs, because an API is how your tools connect to each other. When your calendar, your invoicing app, and your AI assistant all "work together," that is APIs doing the carrying. Whether your tools can talk is a business decision, not just a coding one.

Myth 2: "an API is the same as an MCP." Related, but no. An API is built for developers and is custom per app. An MCP (Model Context Protocol, the subject of Lesson 1) is a universal, AI-shaped adapter that usually sits on top of an API so an AI can use it without custom glue.

🔧 API

Built for developers & software.

– Custom per app: its own menu, auth, and formats
– A developer reads the manual to connect each one
– An AI cannot just pick it up and use it
✓ The raw plumbing that makes tools connect

🔌 MCP

Built for AI agents.

✓ One standard protocol, same for all of them
✓ The AI discovers the tools on its own
✓ Usually wraps an existing API underneath
✓ Connect once, no custom glue per app

The key: an API is the waiter, and an MCP is a standard way to put that waiter in front of any AI. They are companions, not competitors. New to MCPs? Start with Lesson 1: What is an MCP →

03 / 05 · Watch it work

Pick an order. Watch the waiter work.

Choose a request below. You will see the exact request your app sends and the response the API sends back, each line annotated in plain English. This runs entirely in your browser; nothing is sent anywhere. The data is canned for the demo.

Pick a request

↗ Request (what your app asks)

↘ Response (what comes back)

💡 The shape on the right (named fields with values) is JSON, the tidy format most APIs answer in. It is just labeled data: a key, then its value. Want to see APIs doing real work for a business? See the Lead Catcher and Quote Bot kits.

04 / 05 · Connect it safely

5 questions before you connect any API.

Connecting your tools through APIs is how the magic happens, but each connection is a door. Before you open one (especially to email, payments, or customer data), ask these. If a vendor cannot answer them, slow down.

1. What can this API actually do? Read-only, or can it change, send, and delete things on your behalf?
2. What is the key, and who holds it? Most APIs need an API key (a password that proves who is asking). Keep it secret, never paste it into random sites.
3. Where does the data go? Does the response stay with you, or pass through a third party along the way?
4. What does it cost, and are there limits? Many APIs charge per request or cap how many you can send. (Any prices you see are illustrative only, approximate, and change.)
5. What happens if it is wrong or down? What is the worst case, and can you undo it?

05 / 05 · Done

You now understand APIs better than most people who use them daily.

You know an API is the waiter that carries a defined request to the kitchen and brings back tidy data, how it differs from an MCP, and you have watched a real request and response, annotated, run entirely in your browser.

Your apps almost certainly have APIs already. The real power is wiring them together into AI that actually does the work: reads the order, calls the right tools, and finishes the job, safely.

We connect your apps' APIs into AI that actually does the work → Have Rabbithole build it

Companion: What is an MCP? →

Day 22 of 30: free, plain-English AI lessons for small business. See the whole track →