ai-document-intelligence

AI Document Intelligence

A Python SDK for document intelligence and ingestion pipelines for LLM and RAG applications.

The SDK provides a unified pipeline for:

• 📄 Document parsing
• 📑 Page-wise text and Markdown extraction
• 📋 Document metadata
• 🧩 Semantic Markdown-aware chunking
• 🧠 Embedding generation
• 🔗 Standardized document, chunk, and embedding models
• ⚡ Support for multiple document input sources
• 🐍 Strongly typed Pydantic models

The SDK is designed to be domain-agnostic and can be used for financial documents, resumes, reports, contracts, and other business documents.

---

Features

📄 Document Parsing

Parse documents using pluggable parser implementations.

Currently supported:

• LiteParse — local/free parsing
• Llama Cloud — premium cloud-based parsing

The parser layer exposes a unified interface so the rest of the pipeline does not depend on a specific parsing provider.

📑 Page-wise Extraction

Parsed documents provide page-level information including:

• Text
• Markdown
• Page number
• Document elements where supported

📋 Document Metadata

Standardized document metadata is exposed through Pydantic models.

document.metadata.page_count