What Is an Open API?
An open (or public) API is a set of API calls a company publishes for its customers. Instead of logging in and clicking buttons, your own app, spreadsheet, or automation tool sends a request and the product does the work — pulling a report, updating a record, or changing a setting.
The Anatomy of a Single API Call
Every standard API call is made up of five core components. Once you can spot them, any company's API documentation starts to make sense.
One verb that says the job — GET to read, POST to create, PUT / PATCH to change, DELETE to remove.
URL / Endpoint
Query Parameters
Accept: application/json
X-API-Version: 2
The hidden envelope paperwork — what kind of data you're sending, what you'd like back, and which version of the API you're built for.
x-api-key: sk_live_51H8f…
Your secure digital passport — proof that the call is allowed. It usually rides along in the headers (older APIs sometimes tuck it into the address), and without it the answer is 401 Unauthorized.
The information you're actually handing over — only used by POST, PUT and PATCH calls. A GET call usually has no body: the verb, the address and your login details are enough.
1. What Type of API Do They Offer?
REST is the most common for business APIs, but companies choose from four main styles depending on their needs.
GET /api/v2/customers?region=EuropeExamples: Stripe, Twilio, Pelican Wireless and most modern software<soap:Body><m:GetTemperature /></soap:Body>Examples: Older banking networks, Salesforce's legacy integrationsPOST /graphql → { profiles(region:"Europe") { name } }Examples: GitHub, Shopify, FacebookPOST /ThermostatService/GetTemperaturewss://live.company.com/thermostatsExamples: Crypto price tickers, ride-share GPS tracking, chat appsGET /thermostats/TestThermostat?value=temperature comes back as 200 OK · {"temperature": 21}. The verb and the thing you want both sit in the address, and the answer is plain enough to read straight off the screen.
One POST carries the whole request as tags: <soap:Body><m:GetTemperature><m:Name>TestThermostat</m:Name></m:GetTemperature></soap:Body>. The reply is more XML, so your code has to open the envelope before it can find the number inside.
POST /graphql with {"query": "{ thermostat(name: 'TestThermostat') { temperature } }"} returns exactly {"data":{"thermostat":{"temperature":21}}} — the one field you asked for, nothing extra.
gRPC runs the named procedure /ThermostatService/GetTemperature with the data as compact binary instead of text. A WebSocket opens a live pipe at wss://live.company.com/thermostats: you send {"type":"getTemp"} once, and new readings are pushed to you as they change.
The short version: REST, GraphQL and SOAP all send text you can read; gRPC and WebSockets go binary to be fast. And the four verbs and the many-doors rules in the next two sections are REST's way of doing things, not every style's.
2. Which Method Does Each Call Use?
The method is the verb at the start of a request — it tells the API what you want to do. These four verbs are how REST works. The other styles hide the action inside the message, so the verb stops carrying meaning.
GETRead data — "show me my thermostats" or "list my orders".POSTCreate something new or send data — "add a customer" or "set the temperature".PUT / PATCHUpdate something that already exists — "change this customer's email".DELETERemove something — "cancel this appointment".The verb is the instruction. GET reads, POST creates, PUT / PATCH changes, DELETE removes.
Nearly every call is sent as a POST. The action you want is named inside the XML message instead.
One address, one POST. The message body says query to read or mutation to change.
Its own fast binary language. The call names the method directly, like ThermostatService / SetTemperature.
One GET opens the pipe. After that there are no verbs — each message carries its own type, like "type": "setTemp".
The short version: REST puts the verb in the address line, the others put it inside the message. That's one reason REST is the easiest style to read in a company's API documentation — you can often tell what a call does before you open it.
3. Where Is the Endpoint?
The endpoint is the exact web address a call is sent to — the door you knock on. A method only tells the API what to do; the endpoint tells it where to do it. Every company puts its doors in different places, and some give you many while others give you only one, so the address always comes from their documentation.
https://demo.officeclimatecontrol.net/api.cgiobject=Thermostat, or /v2/customers?region=Europe&interest=hiking, or selection=name:TestThermostat&value=heatSetting;coolSetting;temperaturehttps://demo.officeclimatecontrol.net/api.cgi?username=…&password=…&request=get&object=Thermostat&selection=name:TestThermostat&value=heatSetting;coolSetting;temperature — the base address, the login, the method, the resource and the fields you want back, all in one line. A REST API splits the same idea across the address and the request body instead of cramming it into the URL.One address per thing — /customers, /customers/42, /orders. The address itself tells you what the call touches.
A single address serves the whole service. Which operation runs is written inside the XML message.
One address for everything. The message body decides what comes back.
The address names a procedure to run, like ThermostatService / SetTemperature — usually written in code rather than typed into a browser.
One address opens the pipe. After that the messages just flow, with no address on each one.
The company sends calls to an address you publish, so their endpoint list is your address.
The short version: only REST gives you a long list of addresses. With GraphQL, SOAP or WebSockets, one address can hide thousands of possible actions — so the list of operations in the documentation tells you more than the address does.
4. How Do You Prove Who You Are?
How you log in (authenticate) depends entirely on how the company built its system. These are the four most common ways.
?username=demo&password=thermostatExample: Pelican Wireless puts your username and password right in the URL — no key or token.x-api-key: sk_live_51H8f7aD2eP9x…Example: Weather services, Stripe, SendGridAuthorization: Bearer eyJhbGciOiJIUzI1NiIs…Example: Most modern apps after a login stepAuthorization: Bearer ya29.a0ARrdaM…Example: Zapier connecting to HubSpot, Google, SlackThe login travels with every single call. Pelican Wireless writes it straight into the address — ?username=demo&password=thermostat. The other way is a header: Authorization: Basic ZGVtbzp0aGVybW9zdGF0 — and that long string is only demo:thermostat scrambled, not encrypted, which is why it must always travel over https://.
You sign into the company's own dashboard, press "Create key", and paste it into your calls: x-api-key: sk_live_51H8f7aD2eP9x…. It keeps working until you delete it, so guard it like a password — and send it in a header, never the address, because addresses get logged and pasted around.
You send your login once to a POST /login address, it hands back a token, and you attach that to every later call: Authorization: Bearer eyJhbGciOiJIUzI1NiIs…. It usually dies after an hour or a day, so you ask for a fresh one — nothing to go and revoke if it leaks.
Three steps, none of them typed by you: your app sends the person to the company's own login page naming the permission it wants (?scope=calendar.readonly), they tick "Allow", and the company sends back a short code your app trades for a token at POST /token. The app never sees the person's password — which is why it's the only style that works for "Connect with Google".
The short version: a password travels on every call, a key is a password you can delete, a token expires on its own, and OAuth is the only style where your app never sees the person's password.
Examples of Companies That Publish API Calls for Customers
Click any company to open its API documentation — the list of calls it lets you make.
Jira
Create issues, search a project and move work through a workflow.
Shopify
Manage products, orders, inventory and customers in your store.
HubSpot
Add and update contacts, deals and companies in your CRM.
Salesforce
Read and write leads, accounts and opportunities.
Slack
Post messages, create channels and read conversations.
Pelican Wireless
Read and change smart thermostat settings and building schedules.
Stripe
Create charges, refunds, customers and subscriptions.
Google Calendar
Create, move and list events on a calendar.
Zoom
Schedule meetings and pull recordings and attendance reports.
GitHub
Manage repositories, issues and pull requests.
Twilio
Send text messages, make phone calls and check delivery.
Figma
Read design files and layers, pull comments, and export artwork.
Where to Go Next
Keep going in order, or jump to any page from the top menu.