Skip to main content

Welcome to Unstract

Unstract is an open source, no-code platform that helps you automate complex business processes that involve long, complicated documents with a human in the loop.

Unstract goes beyond what current IDP (Intelligent Document Processing) and RPA (Robotic Process Automation) systems are capable of achieving by leveraging the latest in AI: Large Language Models.

A No-Code Platform​

The open source, no-code Unstract Platform is a powerful system for sophisticated automation.

img alt

In 3 simple steps, automate complex business processes that involve long, complicated documents with a human in the loop.

  • Step 1: Harness the capabilities of powerful prompt engineering to support custom document types with Prompt Studio, a no-code environment designed for handling complex documents. Utilize a combination of LLMs, Vector DBs, and Embedding models, monitor extraction field fill rates, evaluate prompt success across document samples, and more.

  • Step 2: Connect sources to retrieve unstructured documents from a variety of storage systems, such as object storage (AWS S3, for example) and file management systems (Google Drive, Dropbox, Box, etc.) Additionally, connect data destinations and send processed, structured data to platforms like Snowflake, BigQuery, Redshift, PostgreSQL, and MySQL, among others. Workflows can also be deployed as APIs, allowing you to POST unstructured documents to the API and receive structured JSON data in response.

  • Step 3: Deploy workflows as:

    • Unstructured data APIs: API deployments are most useful when you have an existing application that deals with unstructured document processing or data processes that you want to automate.
    • Unstructured data ETL Pipelines: Unstract's ETL pipelines are essential for use-cases that require unstructured documents to be processed, structured, and delivered to your object storage system, data warehouse or data lake.
    • Custom Q&A Apps against internal libraries: Several of your business processes require less technical users to interact with unstructured data and documents. This option enables you to launch Q&A apps, each with a unique URL. These apps can be secured with Single Sign-On (SSO), a feature available in Unstract Enterprise.

Unstract Open Source​

The Unstract Platform is released as open source software, under the AGPL license. With the open source release, it's possible to leverage the power of the LLM ecosystem (LLMs, Vector DBs and Embedding models) to automate various business processes.

Advanced enterprise features such as LLMChallenge (use a second LLM to achieve output consensus) as well as cost- and latency-reducing features like Single-Pass extraction and Summarized Extraction are only included in the Unstract Enterprise platform. For more details about these features, please see Unstract Cloud Edition.