What an API is, in plain words

An API is a connection that lets one piece of software ask another piece of software to do something. The letters stand for application programming interface. Wikipedia describes it as something that¹ "is a connection between computers or between computer programs".

That's the whole idea. The rest of this page is examples.

A waiter, for the idea only

Think of a café. You don't walk into the kitchen. You tell the waiter what you want, and the waiter brings it back. The waiter is the agreed way to reach the kitchen.

An API plays the waiter's part between programs. One program asks. The other one does the work and sends something back. Neither needs to know how the other is built inside.

A real example: a weather report

¹ Wikipedia gives a weather sensor as its example. Send the sensor the right message and it replies with a weather report. The message is called an API call, and the report is called the response.

A forecast app on your phone works this way. It doesn't measure the weather itself. It asks a weather service through an API and shows you the answer.

Why you rarely see one

You use APIs all day without noticing.¹ The same source says "an API is typically not visible to users". It works under the hood, so you only see the result, such as a map on a booking page or a card payment going through.

A screen you click on is a user interface. An API is the same kind of doorway, but built for software rather than for people.

The agreement part

¹ Wikipedia says an API is "often compared to a contract". The company that offers the API promises how it will behave. The people who build on it rely on that promise.

That's why APIs come with written documentation. It lists what you can ask for, how to ask, and what you'll get back. You'll meet those parts on the next pages: requests and responses, endpoints and methods and JSON.

Web APIs

Most APIs you'll hear about in business are web APIs.¹ These are described as the kind "which allow communication between computers that are joined by the internet". They use the same basic back-and-forth as a web browser does.² MDN names the two halves of that back-and-forth: the client's messages are requests, and the server's answers are responses.

Where AI comes in

AI makers publish APIs too. A business app with an AI feature may send your text to an AI company's API and show you the reply. Chat windows on the makers' own sites are one way to reach a model. An API is another way, built for software.

Makers set rules for that access.³ OpenAI's documentation says its API has rate limits, which restrict how often a user or client can use the service in a set period. There's more in the page on limits and billing.

Keep secrets and customer data out of prompts unless your business's own policy allows them. Your app's settings and the maker's documentation say what gets sent.

Words you'll hear

• Call or request: one message from your software to the API.
• Response: what comes back.
• Documentation: the maker's written guide to the API.
• Integration: joining two services through their APIs.