HTTP Results

The HTTP Results node is used to define the structure of the HTTP response when running a pipeline as an API. It allows the selection of which data lanes (example Text, Audio, Table) should be returned as part of the final HTTP response.

Configuration

• Lanes – Each lane corresponds to a data type that the pipeline can output. You can specify one or more of these to be included in the API response
  • Lane Types
    • Text – Plain text output
    • Audio – Audio content
    • Documents – Structured document output
    • Image – Image data
    • Questions – Query strings or inputs from earlier steps
    • Table – Structured tabular data
    • Video – Video file output
    • Answers – Final outputs typically generated by an LLM or retrieved from a vector store
• Use the “Add Another Folder” button to define multiple target lanes. These are dynamically returned in the JSON response from the pipeline endpoint

Inputs

• The HTTP Results node supports inputs from all major data lanes
  • Answers
  • Audio
  • Documents
  • Image
  • Questions
  • Table
  • Text
  • Video
• Each of these lanes accepts processed data from upstream nodes to be included in the HTTP response

Outputs

• This node has no downstream outputs. It represents the endpoint of an HTTP-exposed pipeline, serving as the final response formatter for API consumers.

Usage Instructions

1. Drag the HTTP Results node into your pipeline canvas
2. Connect any nodes that generate output to the corresponding input ports (example connect a summarizer to Text, or an OCR module to Documents)
3. Use the right-side panel to specify which lanes should be returned
4. Click Save to apply the configuration
5. When the pipeline is executed via API, only the selected lanes will be included in the response payload

Notes

• If a selected lane receives no input, its field may be omitted or empty in the response
• This node is required for API-triggered pipelines that must return structured output over HTTP