100% Local RAG Intelligence

Engine uninitialized. Please load the database and models above to begin.
Offline Mode (No API Calls)
🌐 Welcome to the Local CIA Factbook RAG Explorer!

This demo showcases a complete Retrieval-Augmented Generation (RAG) pipeline running entirely inside your browser. All data processing, vector similarity calculations, and language model text generation are executed client-side on your device CPU or GPU via Transformers.js and ONNX Runtime Web.

Requirements & Tips:
- Download size is around 400MB to 800MB (cached in browser storage after first load).
- ⚡ WebGPU Acceleration: We use the advanced Transformers.js v4 WebGPU runtime. It is highly recommended to choose the WebGPU backend for hardware-accelerated, real-time generation. Running on the GPU offloads all heavy calculations, keeping your browser UI completely smooth and responsive.