{
  "schemaVersion": "1.0",
  "updates": [],
  "companyNews": [],
  "id": "b918",
  "language": "en",
  "name": "Unstructured",
  "originalName": "Unstructured",
  "library": "tools",
  "category": "Model development",
  "categoryKey": "模型开发",
  "categoryKeys": [
    "模型开发"
  ],
  "categories": [
    "Model development"
  ],
  "entryPoints": [],
  "summary": "Open-source Python components partition PDFs, images and office documents into structured elements for LLM retrieval and knowledge pipelines.",
  "introduction": "This entry covers Unstructured’s open-source preprocessing library. File-specific parsers, layout and OCR components convert documents into structured elements for LLM retrieval and knowledge pipelines. It requires a development environment and format-dependent dependencies. The library differs from hosted Transform; hosted model results should not be attributed to it. Default lightweight telemetry can be disabled as documented.",
  "capabilities": [],
  "company": {
    "details": {
      "facts": [
        {
          "key": "website",
          "label": "Company website",
          "value": "https://unstructured.io/"
        },
        {
          "key": "focus",
          "label": "Business focus",
          "value": "Provides document transformation and preprocessing for AI applications, maintaining both open-source libraries and commercial hosted products."
        }
      ],
      "otherProducts": []
    },
    "name": "Unstructured Technologies, Inc.",
    "originalName": "Unstructured Technologies, Inc.",
    "country": {
      "codes": [
        "US"
      ],
      "names": [
        "United States"
      ],
      "display": "United States",
      "basis": "headquarters",
      "basisLabel": "Company headquarters"
    }
  },
  "officialUrl": "https://github.com/Unstructured-IO/unstructured",
  "profileUpdatedOn": "2026-10-21",
  "sources": [
    {
      "kind": "product",
      "url": "https://github.com/Unstructured-IO/unstructured"
    },
    {
      "kind": "documentation",
      "url": "https://github.com/Unstructured-IO/unstructured/blob/main/README.md"
    }
  ],
  "news": [],
  "links": {
    "interactive": "products/unstructured--b918.html",
    "reading": "read/en/products/unstructured--b918.html",
    "markdown": "read/en/products/unstructured--b918.md",
    "json": "read/en/products/unstructured--b918.json"
  }
}