Graphor icon

Graphor

Interact with Graphor API for document processing, RAG, and AI-powered document chat

Actions12

Extraction → Extract Data

AI-generated

Summary

Extract structured data from specified documents guided by natural language instructions and a JSON Schema definition.

Inputs

  • File IDs — Comma-separated list of file IDs to extract from.
  • File Names (Deprecated) — Comma-separated list of file names to extract from; use File IDs instead.
  • User Instruction (required) — Natural language instructions guiding the extraction process.
  • Output Schema (required) — JSON Schema defining the expected structure of the extracted data.
  • Thinking Level — Controls the model and thinking configuration; options are Fast, Balanced, or Accurate.

Output shape

a JSON object matching the requested Output Schema per input item

Returns extracted data structured as defined by the Output Schema. If multiple file IDs or names are provided, extraction applies to all specified files. Errors per item are returned if extraction fails, depending on node configuration for error handling.

Examples

Example 1: Extract invoice information from specific uploaded files

Provide comma-separated File IDs of the invoices, specify User Instruction to extract invoice fields, define Output Schema to match invoice data structure, and select Thinking Level according to extraction complexity.

Links

Discussion