Understanding Data Flow and Variables

Last updated: January 1, 2026

Overview

Understanding how data flows through your workflow and how to reference it is essential for building sophisticated automations. Variables make data available from one step to the next.

Variable Syntax

All variables follow a consistent syntax pattern:

@StepType.OutputFieldName

or

@SpecificStepName.OutputFieldName

Variable Sources

Intake Variables

Data collected in the Intake step is available throughout the flow:

  • Free-text: @Intake.Message

  • Form fields: @Intake.FieldName (e.g., @Intake.CustomerName, @Intake.ContractTerm)

Step Output Variables

Outputs defined in Agent steps are available as:

  • @AgentStepName.OutputFieldName

For example, if you have an Agent step called "RiskAnalysis" with an output "RiskLevel", it's referenced as:

  • @RiskAnalysis.RiskLevel

Collect Step Variables

Information gathered in Collect steps is available as:

  • @CollectStepName.FieldName

Data Flow Examples

Free-Text Flow

  1. Slack mention → message captured as @Intake.Message

  2. 2. Agent step analyzes message → outputs available as @AgentStepName.OutputName

  3. 3. Condition step checks @AgentStepName.RiskLevel

  4. 4. Approval routes document with context from earlier steps

  5. Notify step references @Intake.Message in the notification

Form-Based Flow

  1. Form submission → fields available as @Intake.FieldName

  2. 2. Agent step uses @Intake.CustomerName and other fields for context

  3. 3. Generate step maps @Intake.CustomerName to template fields

  4. 4. Approval routes generated document

  5. 5. Sign step collects signatures on generated document

Using Variables in Configuration

Variables can be referenced in:

  • Agent step instructions (context)

  • Generate step template mapping

  • Notify step messages and conditions

  • Condition step logical expressions

  • Collect step recipient definition

  • Approval routing and conditions

Best Practices

  • Use descriptive field names in Intake forms to make variables clear

  • Document what outputs your Agent steps provide

  • Use variables to pass relevant context between steps

  • Chain steps together using outputs from previous steps

  • Test variable references to ensure they're available where needed