yashsharmaskk/J.A.R.V.I.S ? reverse-engineered prompt

Build me a local JARVIS style AI assistant for Windows that I can run from one click and open in my browser.

I want a clean web interface where I can type or talk to the assistant, see replies stream in real time, and use the microphone button for voice input. If an NVIDIA GPU is available, it should use it for faster responses, but if not, it should still work on CPU without breaking. Please include automatic model download and simple setup scripts so I can get it running quickly on my machine. The app should also expose a basic status page or API endpoint so I can tell whether the model is loaded, whether GPU is being used, and whether voice recognition is working.

Make it feel like a modern assistant, responsive and fast, with sensible fallbacks if speech recognition or GPU support is missing. If you need to check current docs online for any library details, go ahead and do that.