OpenContracts is an open-source, API-first LLM workspace designed for enterprise document analytics. It empowers users to extract, redact, and manage data from unstructured documents, offering a prompt playground and annotation tools for deep document analysis.

How It Works

The platform utilizes a pluggable microservice architecture for document processing, supporting custom parsers, embedders, and thumbnail generators. It features a Django backend with pgvector for hybrid vector storage, enabling the combination of structured metadata and vector embeddings. LlamaIndex integration allows LLMs to query documents using annotated features and vector stores, facilitating intelligent question answering and bulk data extraction.

Quick Start & Requirements

• Install/Run: Docker is recommended for local setup.
• Prerequisites: Docker, Python.
• Docs: MkDocs Site, Quickstart Guide, Basic Walkthrough

Highlighted Details

• API-first design for programmatic access and integration.
• Pluggable architecture for easy extension of document formats and analysis tools.
• Human annotation interface for manual document markup.
• LlamaIndex wrapper for seamless integration with LLM-powered querying.

Maintenance & Community

• Roadmap includes features like LlamaParse integration, benchmarking, enhanced extraction, streaming, and government data integration.
• Acknowledges contributions from AllenAI's PAWLS and NLmatics nlm-ingestor.

Licensing & Compatibility

• License: GPL-3.0.
• Compatibility: GPL-3.0 may impose copyleft restrictions on derivative works, potentially impacting commercial or closed-source integrations.

Limitations & Caveats

Currently supports PDF and text-based formats (plaintext, MD). While planned, support for other office formats (e.g., DOCX, XLSX) will likely rely on external conversion tools to Markdown.