The Simplest Truth About Adding AI to a Workflow

An AI module is just an HTTP request. Your workflow makes an API call to OpenAI, Anthropic, or another LLM provider. It sends a prompt (text in). The LLM processes it. Text comes back out. Then the rest of your workflow handles that text like it would any other data.

That's it. It's not magic. It's the same input-processing-output flow you've been building all along. Except instead of a formatting rule or a conditional branch, the processing is done by a neural network.

Most automation platforms — Make, n8n, Zapier — have native AI modules that hide the HTTP details from you. You don't see the request body or headers. You see a form where you drop in your API key, pick a model, write your prompt, and hit test. The platform handles the API call behind the scenes.