G-AI Desktop Studio

A browser-based prompt engineering studio for Groq's lightning-fast LLMs, developed at the University of Virginia as part of the UVAi Vanguard Project.

Developed by: Dr. Hudson Golino
Initial funding: The Jefferson Trust

Live Demo

Access the app directly: G-AI Desktop Studio

Features

• Real-time AI Chat — Connect to Groq's fast inference API and chat with state-of-the-art LLMs
• Dynamic Model Selection — Automatically fetches and displays all available Groq models
• Prompt Engineering Tools — Built-in templates and guides for effective prompting
• Advanced Parameters — Fine-tune temperature, max tokens, top-p, and streaming options
• File Upload Support — Upload and analyze TXT, PDF, DOCX, CSV, and JSON files
• Session Management — Export, save, and load conversation sessions
• Fully Client-Side — Your API key stays in your browser

Getting Started

1. Get a Groq API Key

   • Sign up at console.groq.com
   • Create a new API key (free tier available)

2. Use the App

   • Open the app in your browser
   • Paste your Groq API key
   • Click "Connect to Groq"
   • Start chatting!

Security & Privacy

• Your API key stays local — It's stored in your browser's localStorage and never sent to any server except Groq's API
• No server-side storage — This is a fully client-side application
• HTTPS encrypted — All API communications are encrypted in transit
• You control your key — Clear your browser data anytime to remove it

Note: As with any browser-based tool, avoid using API keys with unlimited spending on shared or public computers.

Technical Details

Architecture

Browser (GitHub Pages)
    ↓ HTTPS
Cloudflare Worker (CORS Proxy)
    ↓ HTTPS  
Groq API

The app uses a Cloudflare Worker as a CORS proxy since browser security policies prevent direct API calls to Groq from a webpage.

Built With

• Vanilla HTML, CSS, and JavaScript (no build tools required)
• Mammoth.js — DOCX parsing
• PDF.js — PDF parsing
• Cloudflare Workers — CORS proxy

Repository Structure

├── index.html          # Main application (rename from groq_desktop_app-github-pages-v4.html)
└── README.md           # This file

License

This project is open source and available under the GNU V3 License.

Acknowledgments

• Groq for their incredibly fast LLM inference API
• Cloudflare for free Workers hosting
• The Jefferson Trust for initial funding support to Dr. Hudson Golino
• University of Virginia for institutional support