Skip to content

DEMO_REFERENCE.md

Latest commit

 

History

History
213 lines (155 loc) · 6.5 KB

DEMO_REFERENCE.md

File metadata and controls

213 lines (155 loc) · 6.5 KB
Raw

☕ GEISHA GAINS - QUICK REFERENCE CARD

For Demo Day • Keep this handy! 📋

🚨 EMERGENCY COMMANDS

# If app crashes
npm run dev

# If database is corrupted
npm run db:push

# If Prisma client missing
npm run db:generate

# Kill port 3000
lsof -ti:3000 | xargs kill -9

# View database
npm run db:studio

🎬 DEMO CHECKLIST

Before Demo

  • Run npm run dev
  • Open http://localhost:3000/dashboard
  • Verify market prices are updating
  • Check Caffeine meter is charging
  • Test "Analyze Market" button
  • Confirm trade log is visible

During Demo

  • Explain Brutalist design (white/black/red)
  • Show AI confidence scores
  • Point out Caffeine meter
  • Activate Overdrive (the wow moment!)
  • Show [GLITCH] tags in trade log
  • Highlight SVG charts

🎤 TALKING POINTS

"What is it?"

"A Brutalist crypto trading simulator that uses NVIDIA NIM AI to analyze markets and execute trades. We've added a unique 'Caffeine Overdrive' mode that bypasses AI safety checks for aggressive trading."

"What makes it unique?"

"The Caffeine Overdrive system - it's like an über charge in Team Fortress. When activated, the entire dashboard glitches red, and we execute trades without AI confirmation. It's high-risk, high-reward trading on pure signals."

"What tech did you use?"

"Next.js 14, PostgreSQL with Prisma, NVIDIA NIM for Llama-3 inference, and Uphold's API for real-time crypto prices. The UI is pure Brutalism - thick black borders, no rounded corners, SVG charts instead of Chart.js."

"Why Brutalism?"

"Brutalism represents raw, honest design. In finance, that's exactly what you need - no sugar-coating, just data and action. The aesthetic reinforces the app's aggressive trading philosophy."

🎮 DEMO FLOW (3 MINUTES)

0:00-0:30 - Introduction

  • "This is Geisha Gains, a Brutalist crypto trading simulator"
  • "Built for BugsByte 2026 using Uphold + NVIDIA NIM"

0:30-1:30 - Normal Trading

  • Click "ANALYZE MARKET NOW"
  • "NVIDIA's Llama-3 analyzes BTC, ETH, XRP"
  • "See these confidence scores? We only trade when > 75%"
  • "Every trade updates our wallet and P&L in real-time"

1:30-2:30 - Caffeine Overdrive

  • "Now the gimmick - Caffeine Overdrive"
  • "This meter charges from trades and time"
  • Click "ACTIVATE CAFFEINE OVERDRIVE"
  • 🔴 SCREEN GLITCHES RED 🔴
  • "Notice the glitch effect - this is our power-up state"
  • "We're now executing ANY trade, no AI confirmation needed"
  • "See the [GLITCH] tags? Those are overdrive trades"

2:30-3:00 - Technical Wrap-up

  • "Under the hood: Prisma + PostgreSQL + TypeScript"
  • "Real-time AI with NVIDIA NIM streaming"
  • "Pure SVG graphics for the Brutalist aesthetic"
  • "Fully deployable, all the code is on GitHub"

🐛 TROUBLESHOOTING

Problem: Dashboard won't load

Fix: Check console (F12), likely API route error

# Restart server
npm run dev

Problem: No trades executing

Fix: Check if analysis is running

# Check terminal for "NVIDIA NIM analysis" logs
# If no logs, click "ANALYZE MARKET NOW" manually

Problem: Overdrive button disabled

Fix: Meter must be at 100%

# For demo, you can manually update charge in browser console:
# Open React DevTools → Find UberChargeContext → Set charge to 100

Problem: Glitch effects not showing

Fix: CSS may not be loaded

# Check if globals.css has glitch animations
# Refresh page with Ctrl+Shift+R (hard refresh)

Problem: Database errors

Fix: Reinitialize database

npm run db:push

🎨 DESIGN PRINCIPLES

Brutalism Checklist:

  • ✅ Thick black borders (4px) everywhere
  • ✅ Pure black text on white background
  • ✅ Red (#FF0000) for urgency/danger only
  • ✅ No gradients, no shadows (except sharp brutalist shadows)
  • ✅ Uppercase, bold typography
  • ✅ Monospace for numbers
  • ✅ Raw SVG, no fancy chart libraries
  • ✅ Grid-based layouts with visible structure

📊 KEY METRICS TO MENTION

  • 3 Crypto Assets: BTC, ETH, XRP
  • 10-Second Polling: Auto-analysis every 10s
  • 75% Confidence: Threshold for normal trading
  • 5-Minute Overdrive: Duration of power-up mode
  • 4 Database Tables: User, Wallet, Transaction, MarketSnap
  • 2 API Endpoints: Analyze + History

🎯 JUDGE QUESTIONS (PREP)

"Why did you choose this stack?"

"Next.js for full-stack capability, Prisma for type-safe database access, NVIDIA NIM because we wanted real AI - not just buzzwords - and Uphold for reliable market data."

"What's the hardest part you built?"

"The SVG chart system. We didn't use Chart.js - we built raw SVG paths to fit the Brutalist aesthetic. Also, syncing the Overdrive state across all components in real-time was interesting."

"Is this production-ready?"

"Yes! We're using PostgreSQL, have proper error handling, fallbacks for API failures, and it's deployable to Vercel right now."

"What would you add next?"

"WebSocket connections for sub-second price updates, more trading pairs, historical backtesting, and maybe a mobile app with the same Brutalist design."

"How does the AI actually work?"

"We send current prices and 24h changes to NVIDIA NIM's Llama-3 model. It analyzes momentum, volatility, and market sentiment, then returns BUY/SELL/HOLD with a confidence score and reasoning."

💡 DEMO TIPS

✅ Practice the Overdrive activation - timing is everything
✅ Have backup .env - in case variables get corrupted
✅ Pre-charge the meter - so you can activate on demand
✅ Clear browser cache - avoid stale CSS issues
✅ Test on actual demo machine - not just your laptop
✅ Have Prisma Studio open - show database if asked
✅ Know your numbers - how much P&L, how many trades, etc.

🔗 USEFUL LINKS

🏆 WINNING STRATEGY

  1. Lead with the gimmick - Caffeine Overdrive is memorable
  2. Show technical depth - Prisma schema, NVIDIA NIM, TypeScript
  3. Emphasize design - Brutalism makes you stand out
  4. Live demo the glitch - Visual effects impress judges
  5. Know your code - Be ready to explain architecture

YOU GOT THIS! ☕🏆

Built by Coffee Driven Development
BugsByte 2026

Print this. Keep it at your demo station. Good luck!