unstract: a platform for turning unstructured documents into structured JSON using natural language prompts

What it solves

Unstract automates the process of turning unstructured documents—such as PDFs, images, and scans—into structured JSON data. It replaces the need for writing complex regular expressions or building custom templates for every different document vendor, allowing users to define extraction schemas using natural language prompts.

How it works

The platform uses Large Language Models (LLMs) to parse documents. Users define what they want to extract via a "Prompt Studio," and the system processes the files through a pipeline of text extractors (like LLMWhisperer or Unstructured.io) and LLM providers (such as OpenAI, Anthropic, or Ollama). The resulting structured data can be deployed as a REST API or integrated into an ETL pipeline that moves data from sources like S3 or Google Drive into data warehouses like Snowflake or BigQuery.

Who it’s for

It is designed for teams in data-heavy industries such as finance, insurance, healthcare, and KYC/compliance who need to extract specific information from a wide variety of document formats.

Highlights

• Prompt Studio: Define extraction schemas using natural language instead of code.
• Multi-Provider Support: Compatible with a wide range of LLM providers (OpenAI, Anthropic, Bedrock, Gemini, Mistral, Ollama) and vector databases (Qdrant, Pinecone, Weaviate).
• Extensible Integration: Includes an MCP server for AI agents, an n8n node for automation workflows, and a broad set of ETL connectors.
• Broad Format Support: Handles PDFs, DOCX, spreadsheets, presentations, and various image formats.
• Enterprise Features: Offers dual-LLM verification (LLMChallenge), human-in-the-loop review, and SOC 2/HIPAA compliance in its managed version.