How to use LLM nodes in workflows

To leverage AI capabilities within your Sparkflows workflows, you can use the Interactive LLM Agent node. This node enables direct interaction with a configured Large Language Model (LLM) as part of a workflow.

Step-by-Step Guide

  1. Add the Interactive LLM Agent Node

    • From the Node Library, drag and drop the Interactive LLM Agent onto the workflow canvas.
  2. Select GenAI Connection

    • Open the node configuration.

    • In the Select Connection field, choose your GenAI connection.

    • If the connection has already been configured via the Admin Panel → Administration, it will automatically be available in this dropdown.

  3. Configure Metadata / Content Column

    • Specify the Metadata Column that contains the content you want to pass as input to the LLM.

    • This column typically holds text such as documents, descriptions, logs, or any other unstructured content that the model should process.

    • The selected column’s values are sent to the model along with the prompt during execution.

  4. Define the Prompt

    • In the Prompt section, enter the instruction or query you want the LLM to act on.

    • The prompt can reference the content from the metadata column to guide the model’s response.

    • Prompts should be clear and task-oriented to ensure consistent and accurate outputs.