LangExtract
Python library for pulling structured, source-linked data out of text with LLMs
What it is
LangExtract takes a short prompt and a few worked examples, then uses an LLM to extract entities and attributes from long documents, recording the exact character span each value came from. It chunks large texts, runs passes in parallel, and writes an interactive HTML file for reviewing results in context. Developers and researchers working with clinical notes, reports, or literature use it with Gemini, OpenAI models, or local models through Ollama.