presentsUNAIVERSE
The UN AI timeline

Learn AI /AI Agents and Automation /API

APIAPI

Also called: application programming interface.

Detail

In five seconds

What it is

An application programming interface defines rules that software uses to request data or actions from another system.

In plain words

Think of it as a formal service desk for software. One program asks for something in a standard way, and another program returns an answer in a standard way.

An everyday example

Amina goes to a busy takeaway counter in Nairobi and fills in a small order slip: one lentil soup, no chilli, pay by card. She hands the slip to the staff member at the counter instead of walking into the kitchen. The kitchen follows the standard menu and returns either the meal or a note saying an item is unavailable. Because every slip uses the same format, the kitchen can handle many orders quickly and the cashier can track what happened.

Where the story stops working

A real API usually works machine to machine, not person to person, and responses are often data rather than physical items. Also, an API can expose many different actions and rules, while a takeaway menu is usually simpler. The analogy also hides security details such as keys, authentication, rate limits and error codes.

Remember thisAn API is an agreed interface through which one system requests data or an action from another, using defined operations, inputs, authentication rules and structured responses.

Understand it

Understand it

Why it matters

If you use digital services, you are already affected by APIs even if you never see them. They are the links that let systems share data and trigger actions. For non-specialists, this matters because APIs shape what can be automated, what data moves where, who has access, how errors are handled, and what oversight is needed. In AI systems, APIs often determine whether a tool only chats or can actually act in other systems. Knowing this helps you ask practical questions: What system is being called? What data leaves our environment? What permissions are granted? What happens if the response is wrong or delayed? Who reviews the result before action?

How an API request moves from one system to another

After viewing this, a reader should be able to explain that an API is a structured request-and-response path between software systems, with checks and a predictable output.

  1. 01

    1. Calling app

    A program needs data or wants another system to do something, such as translate text or fetch a record.

    Draw a box labelled 'Calling application' with a small note underneath: 'Needs data or action'.

  2. 02

    2. Structured request

    The program sends a request in an agreed format. It may include text, parameters, and proof that it is allowed to ask.

    Draw an arrow to a second box labelled 'API request' with pill labels: 'action', 'parameters', 'authorisation'.

  3. 03

    3. Service checks and acts

    The receiving service validates the request, checks permissions, runs the task, and prepares an answer.

    Draw a larger middle box labelled 'Service behind API' with two stacked mini-boxes inside: 'check request' and 'perform task'.

  4. 04

    4. Response

    The service returns structured output, such as translated text, a list of records, or an error message.

    Draw an arrow to a final box labelled 'Response' with two branches inside: 'result data' and 'error'.

A left-to-right flow diagram with four boxes linked by arrows. Read from the calling application, to the API request, to the service that checks and performs the task, to the response returned. Small labels show that permission checks and formatting rules sit in the middle, not at the start or end.

On hover or tap, each box highlights and reveals one short example: 'translate paragraph', 'get weather', 'create calendar event', or 'permission denied'.

Describe this without the diagram

The diagram is a four-step flow from left to right. First, a calling application needs data or an action. Second, it sends an API request in a standard format with details such as the action, parameters and authorisation. Third, the service behind the API checks the request and performs the task if allowed. Fourth, it returns a structured response, either result data or an error. The main point is that an API is a predictable pathway between software systems, not a person-facing screen.

With reduced motion: A static four-box flow chart with arrows: Calling application, API request, Service behind API, Response. Example labels appear as fixed text below each box instead of hover effects.

An API lets one piece of software interact with another without using the full screen that a person would use. One system sends a request. The request follows agreed rules about what can be asked for, how to ask, and what format to use.

What goes in is usually a request with details such as an address, a question, a file name, or a command like “translate this text” or “create a calendar entry”. What happens next depends on the service behind the API. It checks the request, may verify permission, does the task, and prepares a response.

What comes out is usually structured data, often in a machine-friendly format. That might be a weather reading, a list of matching records, a translated sentence, or a message saying the request failed. Because the format is predictable, other software can use the result automatically.

This matters in AI because assistants and agents often do not only generate text. They may also call APIs to search a database, send a message, retrieve a document, or trigger another tool. An API is not the same as a connector or an endpoint: a connector is often a ready-made bridge built on top of APIs, while an endpoint is one specific address inside an API.

How “Ordering from a takeaway counter” maps onto the real thing

In the storyIn the AI system
The order slip with fixed fieldsA request with a defined format
The staff member receiving the slipThe service that accepts and checks the request
The kitchen preparing the mealThe underlying system doing the work
The meal or unavailable note returnedThe structured response or error message

A worked example

A programme officer at Headquarters is finalising a side-event concept note for circulation by close of business. The event-planning tool uses the approved room-booking API to check whether a suitable room is available and place a provisional hold.

Starting from: Request: operation=check_availability, capacity=medium, equipment=videoconference, time_window=afternoon

  1. The planning tool sends the request to the room-booking endpoint with its authentication data and the required parameters.
  2. The booking service validates the credentials, request format and the officer's permission to place a hold.
  3. The service checks its booking records and applies the rules for room capacity, equipment and availability.
  4. It returns a structured response with availability, permitted next actions and status information. If the request is valid and a room is available, the tool sends a second request for a provisional hold.

Result: The tool records the provisional hold and inserts the room details into the draft concept note. If no suitable room is available, it displays the service's response rather than guessing or treating an error as a booking.

In a United Nations workplace

During a Resident Coordinator's Office town hall, a chief of section receives an unanticipated question about delivery against the current workplan. An approved internal dashboard calls the results system through its API and retrieves the latest validated status for use in the chief's talking points before the question period closes.

The dashboard is the client. It sends an authenticated request with the relevant workplan identifier to a specified endpoint. The results system validates the request, checks access rights and returns structured status information that the dashboard can display.

Before you try thisThe API must return only information the user is authorised to access. Unpublished reporting, personal data or authentication credentials must not be sent to an unapproved service or exposed in prompts, talking points or system logs.

Where you may hear it

  • A vendor demo for an ICT governance board
  • A procurement meeting on connecting a new platform to existing systems
  • An automation review with a programme office and information management officer
  • A technical discussion before an internal tool is cleared for use

Avoid confusion

Avoid confusion

Often confused with

Common misconceptions

  • Common misconception: An API is a kind of AI.

    Correction: An API is not AI by itself. It is an interface for software-to-software interaction. An AI system may be available through an API, but many APIs simply return database records, maps, payments or other non-AI services.

  • Common misconception: An API is the same as a user interface on a website or app.

    Correction: A user interface is built for people to click, read and type into. An API is built for software to send structured requests and receive structured responses. They may connect to the same backend system, but they serve different users and formats.

  • Common misconception: If two systems have APIs, they will work together automatically.

    Correction: APIs make integration possible, not automatic. Systems still need matching formats, permissions, error handling, rate-limit planning, testing and maintenance. A mismatch in any of these can stop the integration or make it unreliable.

Quick check

Which description best matches an API?

Go deeper

Go deeper

The simple version

Think of it as a formal service desk for software. One program asks for something in a standard way, and another program returns an answer in a standard way.

The technical version

An API defines how a client application can interact with a service through specified operations, inputs, outputs and rules. A client sends a request to a service interface, often to a particular endpoint, with parameters and authentication data. The service validates the request, applies business logic or calls underlying resources, then returns a structured response and status information. In practice, APIs are used to expose functions such as search, translation, storage, payments or messaging so other systems can invoke them programmatically. In AI workflows, an assistant or agent may use APIs as tools to fetch context, perform actions or update external systems. This is related to function calling and tool use, but not identical: function calling usually refers to the model producing a structured instruction for a tool, while the actual API call is then made by software around the model.

Reviewed external resources

  • Primary sourceModel Context Protocol (opens in a new tab)Model Context Protocol

    This link describes Model Context Protocol, a specific way tools and models can connect, and it suits technical readers comparing protocols with general APIs.

    Deeperlink checked 2026-07-31

  • EncyclopaediaAPI (opens in a new tab)Wikipedia

    This link gives a broad overview of APIs, their history and common uses, and it suits readers who want a general reference.

    Intermediatelink checked 2026-07-31

  • ArticleBuilding effective agents (opens in a new tab)Anthropic

    This link explains how AI agents use tools and external systems, and it suits readers connecting APIs to agent behaviour.

    Intermediatelink checked 2026-07-31

  • VideoVideo explanations: API (opens in a new tab)YouTube search

    This search results page helps readers find different video explanations of APIs, and it suits people who prefer to learn by watching because it is not one single source.

    search results, not one pagelink checked 2026-07-31

  • DiscussionCommunity discussion: API (opens in a new tab)Reddit search

    This search results page shows community discussions and practical questions about APIs, and it suits readers who want informal examples and debates because it is not one single source.

    search results, not one pagelink checked 2026-07-31

Where this came from

Drafted with a language model through an offline pipeline, checked against this site's schema, and every external link above fetched to see whether it answers. Treat it as a starting point and check anything that matters against the primary sources. How this site is built and what that means.

The one thing to remember

An API is an agreed interface through which one system requests data or an action from another, using defined operations, inputs, authentication rules and structured responses.