unstructured
Convert complex documents into structured data for language models.
Open-source pre-processing and ETL tools for ingesting complex documents into LLMs and vector stores.
Unstructured is an open-source ETL library that ingests files like PDFs, Word documents, and images, transforming them into clean, structured formats. It suits compliance and GRC professionals needing to parse complex enterprise documentation for downstream machine learning workflows.
What it is
The library provides modular parsing, partitioning, chunking, and embedding functions for over 60 file formats. It supports integration with LLM agents through an MCP server and handles heavy multi-modal documents via containerized or local Python environments.
Who it's for
It suits data engineers and technical compliance officers building automated document ingestion pipelines for LLMs. It does not suit non-technical compliance staff who require a standard point-and-click user interface without code or terminal interactions.
Key features
Multi-Format Partitioning
Parse PDFs, Word docs, emails, and images into clean structured outputs.
MCP Server Support
Bring document processing directly to your agents via Model Context Protocol.
Flexible Installation
Install via PyPI with selective extras or run entirely inside a Docker container.
Developer Tooling
Manage dependencies locally using uv and Makefile development targets.
Up and running in 3 steps
Our honest verdict
What's great
- Supports over 60 distinct file and document types out of the box.
- Includes a Model Context Protocol server for agentic document ingestion workflows.
- Provides pre-built multi-architecture Docker containers for isolated execution.
- Modular design allows installing only the document-type dependencies you actually need.
Where it falls short
- Demands significant setup effort, requiring multiple system-level dependencies like tesseract-ocr, poppler-utils, and libmagic-dev.
- Python environment management and package extra configurations can become complex to maintain.
- Advanced enterprise features and lower-code UI pipelines require shifting to their paid platform product.
Pricing & license
The library is free and open-source under its license, with the only cost being your own compute and hosting infrastructure. Unstructured also offers a paid enterprise platform product for production-grade workflows and a low-code UI.
Open remote AI-training roles
876 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
What document formats does Unstructured support?
It supports over 60 file types including PDFs, HTML, XML, JSON, emails, MS Office docs, and various image formats.
Can I run this tool inside a container?
Yes, you can pull multi-platform Docker images directly or build your own using the provided Makefile and Dockerfile.
