How automation workflows and AI steps work
Visual automation builders like Zapier, Make and n8n all work from the same few parts: a trigger, steps, data passed between them, and decisions along the way. These pages explain each part in plain words, using what the makers' own documentation says. They also show where an AI step fits and where a person should check it.
All the build notes
- What an automation workflow is
What a tool means when it says it can automate a task, and the parts inside one workflow: a trigger, steps and the data between them.
- Triggers: what starts a workflow
How a workflow knows when to start, the difference between polling and instant triggers, and what the makers say about schedules and webhooks.
- Actions and steps: what a workflow does
What a step can do in a workflow, and how each step receives the last step's output and hands its own result to the next.
- What an AI step does inside a workflow
What goes into an AI step, what comes out, and what the language model is actually doing, with notes on cost counting and keeping data private.
- Mapping data between steps
How a name or amount from one step ends up in the next step's field, and the usual ways a workflow breaks when fields don't match.
- Filters, branches and paths
How workflows make yes-or-no decisions with filters, paths, routers and If nodes, and where a person can check an AI answer before it goes on.
- Zapier, Make and n8n: how each describes its building blocks
The words Zapier, Make and n8n use for the same ideas, and how each maker's own documentation describes hosting and how usage is counted.