DocuProx icon

DocuProx

AI-powered document data extraction using DocuProx. Use AI Agents with natural language prompts or AI-driven template extraction to pull structured data from documents in real-time or via high-volume batch processing.

Actions5

Document → Process Agent

AI-generated

Summary

Process a single document image with an AI Agent via the DocuProx API to extract structured data based on natural language prompts or supplied JSON payloads without needing a predefined template.

Inputs

  • Image Source (required) — Choose whether to upload an image file from binary data or provide a Base64 encoded image string.
  • Binary Property Name (required) — Specify the name of the binary property containing the image file when uploading an image.
  • Base64 Image (required) — Provide the Base64 encoded image data if using the Base64 string option.
  • Selection Method (required) — Select how to provide the agent payload: either structured fields or manual raw JSON.
  • Document Type (required) — Specify the type or category of the document to help guide AI extraction (required in structured mode).
  • Custom Instructions — Optional natural language guidance for the AI Agent to tailor extraction.
  • Prompts — Define individual fields and instructions for the AI Agent to extract from the document image (structured mode).
  • Static Values — Add static key-value metadata pairs to include alongside the AI Agent's extraction results (structured mode).
  • Payload (required) — Provide the entire agent payload as a raw JSON object if using manual JSON mode.

Output shape

a list of items each containing a JSON object with the process response from the DocuProx API, including success status, extracted data, and timestamp

The output JSON includes the AI extraction response from the /process-agent endpoint. Errors are thrown for invalid inputs or on failed API calls unless 'continue on fail' is enabled.

Examples

Example 1: Workflow includes an image file from a previous node (in binary property 'data'), uses 'structured' mode with custom fields and instructions, and sends this image and payload to DocuProx API for AI data extraction.

Set 'Image Source' to 'Upload Image File', specify 'Binary Property Name' as 'data', select 'Selection Method' as 'Structured (Form)', fill in 'Document Type', 'Prompts' with keys and instructions, optional 'Custom Instructions' and 'Static Values', then run the node.

Example 2: Workflow provides the document image as a Base64 string and supplies the entire AI Agent payload as JSON manually.

Set 'Image Source' to 'Base64 String', paste the Base64 image data into 'Base64 Image', choose 'Selection Method' as 'Manual (JSON)', and enter the complete AI payload JSON into 'Payload'.

Links

Discussion