Unstructured Transform operations quickstart

Use Python or cURL to turn local files into AI-ready JSON.

This quickstart uses Unstructured Transform to turn your locally hosted source files into AI-ready JSON.

To use Unstructured Pipelines to work with your locally hosted files instead, skip this quickstart and use the Unstructured Pipelines operations quickstart. Pipelines is sold separately from Transform. To purchase Pipelines, contact Unstructured Sales.

To use Pipelines to work with remotely hosted files, skip this quickstart and use the Unstructured Pipelines operations remote files quickstart.

1. Get your Unstructured API URL and API key

   Get your Unstructured API URL and API Key by signing into your Transform account:

   1. Go to https://transform.unstructured.io and sign in. If you are new, click Register.
   2. Click API Keys.
   3. Copy API URL and paste this value into a secure location. Do the same with API Key.
2. Create environment variables

   Create these two local environment variables:

   • UNSTRUCTURED_API_URL, and set it to your API URL.
   • UNSTRUCTURED_API_KEY, and set it to your API key.

   To learn how to set these environment variables accordingly, see your system’s documentation.

3. Install required software

   Follow one of these procedures, depending on whether you are using Python or cURL.

   Python SDK

   1. Make sure you have Python 3.11 or higher installed locally. Check your version or install Python.

   2. (Optional) Use or set up a Python virtual environment locally.

      A virtual environment is a per-project Python sandbox. It isolates a project’s Python interpreter and packages from your other projects and system-wide Python setup. It is recommended but not required.

      Popular options include uv (recommended), poetry, venv, and conda.

   3. Install or upgrade the Python SDK locally. (If you are using a virtual environment, replace pip install and pip show as follows with your provider’s related command.)

      To install, run this command:

      pip install "unstructured-client>=0.46.2"
      

      If you already have it, upgrade it with this command:

      pip install --upgrade "unstructured-client>=0.46.2"
      

      To check if you already have it, run this command:

      pip show unstructured-client
      

   cURL

   Make sure you have cURL and jq installed locally.

4. Pick a use case to explore.

   Partition (recommended for starters)

   Turn your documents’ contents into structured text and metadata, which Unstructured outputs in a predefined, consistent, expressive, and contextualized JSON format, ready for AI.

   Extract

   Extract only the data that you care about from your documents, which Unstructured outputs as JSON object data records in a format that you define.

   Enrich

   Uncover more context and insights from your documents, which Unstructured adds to your existing AI-ready JSON.

   Embed

   Power up semantic search and retrieval with vector embeddings that Unstructured generates. These embeddings are based on your documents’ text content.