Skip to main content

AI

The AI step in Snappit allows you to call a Language Model (LLM) using a user-defined prompt and configuration. This step is ideal for generating dynamic outputs, extracting information, or performing reasoning tasks directly within a workflow.

Snappit AI Step

🎯 Purpose​

To invoke an LLM with a custom prompt and parameters, then return structured or unstructured output depending on the configuration.

🧭 Behavior​

  • Sends a prompt to an LLM based on the provided environment settings
  • Accepts an execution environment which can be local or remote
  • Supports two response types: raw text or structured JSON
  • If using the JSON type, validates the response against the provided output schema
  • Makes the LLM response available for use in later steps

⚙️ Configuration​

  • prompt: The text prompt to send to the LLM
  • environment: Specifies where the model executes, either local or remote
  • type: Defines the expected response format, either text or json
  • outputSchema (optional): A JSON schema used to validate the LLM response if the type is json

✅ Use Cases​

  • Generating dynamic responses for customer communication
  • Summarizing content or rephrasing inputs in workflows
  • Extracting structured information using intelligent prompting
  • Performing decision-making or branching logic based on LLM output

The AI step enhances your workflow with flexible, intelligent automation by integrating LLMs directly into the process.