document-extraction

You can extract structured fields from mixed document batches by applying parsing and classification using LandingAI ADE, which outputs hierarchical Markdown and JSON with per-chunk visual grounding. It processes PDFs, images, spreadsheets, and presentations using JSON Schema or Pydantic models.

Yes, async document processing supports large files up to 1 GB or thousands of pages. This capability handles parsing, extraction, and classification for massive documents while providing structured Markdown and traceable grounding metadata.

Yes, you can use Pydantic schemas to extract fields from parsed documents. The extraction capability supports both JSON Schema and Pydantic models, allowing you to define structured field requirements and receive outputs with hierarchical JSON and bounding box grounding.

Visual grounding in document extraction provides bounding boxes, page numbers, and confidence metadata for each extracted field. This traceability ensures every parsed Markdown chunk and JSON result links directly to its source location within the original PDF, image, or spreadsheet.

You can parse and classify multi-document batches by using the split and classify capability, which routes parsed documents for downstream processing. It transforms heterogeneous files into structured Markdown and JSON, identifying document types and extracting fields with layout-aware grounding.

Yes, you need the landingai-ade library to parse documents into structured Markdown. This dependency enables the core parsing, extraction, and classification functions that convert heterogeneous files into layout-aware JSON with per-chunk grounding and confidence metadata.