Unstructured

Ingestion and preprocessing library for RAG, handling PDFs, HTML, and office docs.

Overview

While not a search engine, Unstructured is the critical tool for preparing retrieved documents for LLMs. It parses complex formats (PDF, DOCX, HTML) into clean text elements, preserving hierarchy and structure for better retrieval.

Best for

• Processing PDFs and Office docs
• Cleaning messy HTML for RAG
• Structuring unstructured data

Trade-offs

What XeroHack pre-wires

When you pick Unstructured in the interview, the engine emits these files into your scaffold:

• LangChain UnstructuredLoader
• LlamaIndex UnstructuredReader