INTERACTIVE APP On-Device AI Active

On-Device WebLLM AI Companion & Static RAG

In-browser WebGPU-accelerated language model with static vector Retrieval-Augmented Generation adhering to Carbon for AI component design.

v1.0.0

Engage with our Carbon AI Research Companion. The model runs entirely in your browser using WebGPU and WebAssembly, ensuring complete privacy with zero server telemetry or API calls.

Carbon AI Research Assistant

WebGPU • Static Vector RAG • Local Privacy

SmolLM2-360M-Instruct

How can I assist your research today?

Select a starter prompt or ask any question:

⚡ Local RAG Active AI generated. 100% Client-Side Privacy.

Security & Hardware Acceleration

  • Execution Target: Client WebGPU tensor cores via WebLLM
  • Privacy Model: 100% offline execution after initial weight caching
  • Retrieval Engine: Client-side vector similarity over static rag-index.json
  • Carbon Design System: Fully styled using Carbon AI Chat React component specifications