A guide to building long-term compounding knowledge infrastructure. See details on GitHub .

Agentic Document Extraction

Python library that wraps LandingAI's Agentic Document Extraction API to extract structured data from visually complex documents.

Overview

Agentic Document Extraction is a Python library that simplifies calling LandingAI’s document extraction API and returns hierarchical JSON and ready-to-render Markdown for visually complex documents (tables, images, charts). It supports long PDFs, images, URLs and includes utilities for visualization and grounding.

Key Features

  • Batteries-included installer: pip package with minimal setup.
  • Support for PDFs of any length, images and URLs; automatic splitting and parallel processing for large documents.
  • Structured, hierarchical JSON output plus Markdown rendering.
  • Visualization tools and optional grounding image exports for debugging.
  • Robust retry and backoff handling for API errors and rate limits.

Use Cases

  • Extract structured data from invoices, reports, forms and research papers.
  • Batch processing pipelines to parse large document collections.
  • Build data pipelines that convert scanned documents into searchable records.

Technical Highlights

  • Language: Python (supports 3.9–3.12).
  • Configurable parallelism (BATCH_SIZE, MAX_WORKERS) and retry policies.
  • Connectors for Google Drive, S3, local directories and URLs.
  • Produces both raw bytes and field-extraction outputs with pydantic models.

Comments

Agentic Document Extraction
Resource Info
Author Landing AI
Added Date 2025-09-29
Open Source Since 2025-03-12
Tags
AI Agent Dev Tools Open Source