API (application programming interface)

Most automations, connectors and AI agents that touch your CRM, inbox or ad account are making API calls underneath. You don't need to know how the service works inside, and an AI assistant can read the documentation for you.

How it works

A request goes to a specific web address, called an endpoint, with a method that says what you want: GET to read something, POST to create something. It carries any data the service needs and a credential, such as an API key, that proves which account is asking. The service replies with a status code and usually some JSON, a structured text format that both programs and AI models can read.

The documentation lists each endpoint, the fields it needs and its rate limit, which is how many requests you may send in a given period. Once one call works, you can pass live data from it into an AI step.

Example

An analyst wants yesterday's online orders next to yesterday's ad spend in one sheet. She pastes the orders section of her shop platform's API docs into an AI chat and asks for the single call that returns orders by date. She runs it in her automation tool, gets JSON back, asks the AI to summarize the fields, and maps the order total and date into the sheet.

Common mistake

Firing requests in a loop until the service starts refusing them. Check the rate limit in the docs before you schedule a workflow, and space calls out or batch them so the job finishes without errors.