> ## Documentation Index
> Fetch the complete documentation index at: https://docs.workflowmachine.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Steps

> Understand what steps do inside a workflow and how to combine actions, AI, and logic in sequence.

## What a step is

A step is a unit of work inside a workflow.

After the trigger starts the workflow, each step performs a task. Some steps talk to apps, some transform or inspect data, some use AI, and some control the flow of the workflow itself.

## Different kinds of steps

Workflow Machine supports several common step patterns.

### App action steps

These steps interact with external services.

Examples include:

* sending a message
* creating a page or record
* writing to a spreadsheet
* making an HTTP request

These are often the steps that produce the final user-visible outcome.

### AI steps

AI steps are useful when the workflow needs to work with text or unstructured information.

Common uses include:

* summarization
* extraction
* classification
* drafting
* rewriting

AI works best when the step has a clear job and the expected output is easy to evaluate.

### Control-flow steps

Some steps exist to shape the workflow rather than talk to an app.

Common examples include:

* **If** to branch based on conditions
* **Wait** to pause before continuing
* **End** to stop the workflow early
* state-related steps for reading or setting workflow state

These steps help make the workflow more predictable and efficient.

## Why step order matters

Steps run in sequence, so the output of one step often becomes the input to the next.

That means the order of steps changes the behavior of the workflow. For example:

* summarize first, then notify
* fetch data first, then decide
* check a condition first, then continue only if it passes

Good workflow design usually comes down to putting the right steps in the right order.

## What makes a good step

A good step has:

* one clear job
* only the fields it actually needs
* an output that is useful to the next step

If a step is trying to do too much, the workflow becomes harder to test and troubleshoot.

## A practical design habit

When adding a new step, ask:

* what input does this step need
* what output should it produce
* who or what uses that output next

Those three questions prevent a lot of messy workflow design.

## Start simple

The easiest workflows to maintain are usually made of small, understandable steps. It is better to have a short sequence of clear steps than one confusing chain where the logic is hard to explain.