enoch3712/ExtractThinker

Health v2 · maintenance only

Adoption not part of the score

1595 stars · 154 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

ExtractThinker is a Python document intelligence library that uses LLMs to extract and classify structured data from documents like PDFs, images, and spreadsheets. It provides ORM-style interaction with Pydantic contracts and pluggable document loaders (Tesseract OCR, Azure Form Recognizer, AWS Textract, Google Document AI).

Use cases

• extract structured fields from invoices and receipts with LLMs
• convert pdf documents to structured using pydantic contracts
• classify documents or document sections automatically
• process large multi-page pdfs page by page with splitting strategies
• run ocr on scanned documents and feed results to an llm
• build document processing pipelines with multiple cloud ocr providers
• extract data from spreadsheets and images into typed python objects

When to choose

• you need LLM-powered structured extraction from documents with typed contracts
• you want to swap between OCR providers (Azure, AWS Textract, Google Document AI, Tesseract) behind one interface
• you prefer an ORM-like API for document workflows with async support
• you need document classification and splitting strategies for large files

When to avoid

• you need simple deterministic pdf text extraction without LLMs
• you want a ready-made GUI application rather than a Python library
• your documents contain highly sensitive data you cannot send to external LLM providers
• you need a fully managed document intelligence service with SLAs

Facets

library · maturity active

ocr nlp llm-inference rag pdf parser machine-learning data-science artificial-intelligence large-language-models pdf data-science developer-tools python cross-platform document-intelligence document-processing pydantic-contracts structured-extraction langchain-inspired document-classification invoice-parsing natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="enoch3712/ExtractThinker")