LangExtract — LLM-Powered Structured Extraction

Google open-source Python library (34.7k ⭐) for extracting structured information from unstructured text with source grounding, interactive visualization, and support for Gemini, OpenAI, Ollama, and custom providers.

When to use this skill

• Extract entities, relationships, or attributes from clinical notes, reports, or any text
• Map every extraction back to its exact location in the source (source grounding)
• Process long documents via automatic chunking + parallel LLM calls
• Visualize thousands of extractions in an interactive, self-contained HTML file
• Run local inference without API keys using Ollama
• Define custom extraction schemas without fine-tuning — just a prompt + few examples

Instructions