This repository provides a complete starter kit for building document question-and-answer applications that run entirely locally at no cost. It combines Ollama for local LLM inference, Supabase pgvector for vector storage and similarity search, and Langchain.js for orchestrating LLM interactions. Embeddings are generated using Transformers.js with the all-MiniLM-L6-v2 model, and the frontend is built with Next.js. The system indexes documents locally and retrieves relevant context to power conversational AI responses.