Variable Referencing

Variables are a key component of AirOps Workflows, allowing you to create dynamic, personalized, and contextual content. Learn how to reference variables in your Workflows, whether they come from inputs or from step outputs.

Types of Variables

In AirOps Workflows, you can reference two primary types of variables:

1. Input Variables: Values defined in the Start step of your Workflow, including:

   • Custom inputs (text, numbers, JSON, files, etc.)

   • Brand Kit attributes

2. Step Output Variables: Results generated by previous steps in your Workflow

Referencing Input Variables

Input variables are referenced using the Variable Selector (the pink icon on the top right of an input field) or using Liquid syntax with double curly braces: {{ variable_name }}

Example: Custom Input Variables

When you define an input in your Workflow's Start step (like "topic" or "keyword"), you can reference it in subsequent steps:

{{ topic }}
{{ keyword }}

Example: Brand Kit Variables

If you've connected a Brand Kit to your Workflow, you can reference its attributes:

{{ brand_kit.brand_name }}
{{ brand_kit.brand_about }}
{{ brand_kit.writing_persona }}
{{ brand_kit.brand_url }}

Referencing Step Output Variables

Every step in your Workflow generates an output that can be referenced in subsequent steps. These are referenced using the step number and the output property:

{{ step_1.output }}

For steps that return structured data (like JSON), you can access specific properties:

{{ step_2.output.title }}
{{ step_3.output.keywords[0] }}

Best Practices

• Use the Variable Selector (purple pill button) to easily add variable references without typing the syntax

• Preview outputs in the Test panel to understand their structure before referencing

• For JSON outputs, check the structure in the Test panel to correctly reference nested properties

• Remember that variables can only reference previous steps, not future ones

Common Use Cases

• Pass user input into an LLM prompt: Write an article about {{ topic }}

• Include brand voice in prompts: {{ brand_kit.writing_persona }}

• Process the output of an LLM in a subsequent step: {{ step_1.output }}

• Extract specific data from API responses: {{ step_3.output.results[0].url }}

• Format text with previous step results: The generated headline is: {{ step_2.output.headline }}

By effectively leveraging variables in your Workflows, you can create powerful, dynamic applications that respond to inputs and build upon the results of each step.