Skip to main content
The Generate Text action allows you to generate text dynamically during your endpoint execution using your connected ChatGPT integration. At runtime, this block sends your prompt to the selected model and stores the generated output for use in later blocks.
Generate Text action block

​
ChatGPT Integration

This action requires a connected ChatGPT integration. If no integration is connected, you will be prompted to connect one before the block can run.

​
Prompt

The Prompt field is the instruction sent to the model. You can:
  • Write plain text instructions
  • Include dynamic variables using {} syntax
  • Combine static and dynamic values

​
Example

Write a short welcome email for {params.partner}.
The model receives the fully resolved prompt at runtime.

​
Model

The Model dropdown defines which model generates the response. Different models can vary in:
  • Speed
  • Cost
  • Output quality
  • Context length
Choose the model that best fits your use case.

​
Temperature

The Temperature value controls how deterministic or creative the output is.
  • 0.0 → Very deterministic and predictable
  • 0.5 → Balanced
  • 1.0 → More creative and varied
Lower values are ideal for structured or factual outputs. Higher values are better for brainstorming or creative writing.

​
Reference Variable

The Reference Variable defines where the generated output will be stored. Once set, you can reference the response later in your flow using: 
{local.VARIABLE_NAME}

​
Example

If:
  • Reference Variable: summary
You can later use: 
{local.summary}

​
Behavior & Scope

  • Text is generated at runtime when the block executes.
  • Output is stored only for the current endpoint execution.
  • If another block later uses the same reference variable name, the previous value is overridden.
The Generate Text block is useful for dynamic emails, summaries, classifications, and other AI-powered outputs in your API flow.