Google open-sources LangExtract: free LLM document extraction to replace $100K enterprise tools
mdancho84 · x · 2026-10-28
Google released LangExtract, an open-source Python library that uses LLMs to pull structured data from unstructured text — a free alternative to enterprise extraction tools that cost up to $100K.
Key features:
- Schema-consistent extractions driven by user-defined instructions and few-shot examples
- Precise source grounding: every extracted field maps to its exact location in the source text, with visual highlighting for verification
- Optimized for long documents (100+ pages)
- Generates interactive HTML for human review of extractions
- Works with Gemini, OpenAI models, and local LLMs via Ollama
Available on PyPI (v1.1.1, 24k+ GitHub stars), with examples covering clinical notes, medication extraction, and literary entity extraction.
More from coding & agent
- LangChain lays out the 5 components of a "company harness" for internal agents — hwchase17 · 2026-10-28
- Doubao's Record-to-Skill Feature Automates Apple Notes Migration to Obsidian in Two Minutes — songguoxiansen · 2026-10-28
- One month building a co-op vet game with Claude, Unity, Blender and Meshy — PixelPawsInt · 2026-10-28
- Six AI agents ran a company for months — until a laptop lid slept it for 26 hours — Icy_Comfort_6220 · 2026-10-28
- .NET Conf 2026 agenda live: C# union types, WinUI + AI agents among 34 sessions — unixterminal · 2026-10-28
- After interviewing ~2 dozen teams, researcher says fancy coding agent workflows mostly don't work — thegautamkamath · 2026-10-28