Skip to main content

Introduction

Welcome to the documentation for OpenGIN Tracer.

OpenGIN Tracer is an advanced data extraction pipeline designed to process PDF documents and extract structured table data using Generative AI (Gemini).

What is OpenGIN Tracer?​

OpenGIN Tracer automates the tedious task of manually copying data from PDF tables. It leverages a multi-agent architecture to:

  1. Scan PDFs and split them into manageable pages.
  2. Extract data from each page using Gemini 2.0 Flash.
  3. Aggregate the extracted data across all pages.
  4. Export the final results into structured formats like JSON or CSV.

Key Features​

  • GenAI-Powered Extraction: Uses state-of-the-art LLMs to understand complex table structures.
  • Metadata Extraction: Define custom schemas to extract table-level metadata (e.g., categories, confidence scores).
  • Multi-Agent Architecture: Modular design for better scalability and maintenance.
  • Automated Workflow: End-to-end automation from PDF input to structured output.
  • Extensible: easy to add new agents or extraction logic.

Getting Started​

Check out the Usage section to get started with the CLI, or explore the Architecture to understand how it works under the hood.