Stop overpaying for thrift finds. Upload a video, get instant price analysis powered by AI.

When thrifting or shopping secondhand, it's hard to know if you're getting a good deal. Is that laptop worth $250? What about that vintage jacket? Most people overpay because they don't know the market value.

ThriftSmart AI analyzes product videos in real-time, searches live marketplace data, and tells you if it's a COP (good deal) or DROP (overpriced) — complete with AI-powered voice analysis.

• 📹 Video Product Recognition - Upload any product video
• 🔍 Live Market Research - Scrapes eBay for real-time comparable prices
• 🧠 AI-Powered Analysis - Uses Google Gemini to identify products and analyze deals
• 🎙️ Voice Response - Get spoken recommendations via ElevenLabs TTS
• ⚡ Instant Verdict - Clear COP/DROP decision with detailed reasoning
• 🎨 Modern UI - Beautiful glassmorphism design with smooth animations

Input: Video of HP EliteBook 840 G5 laptop, listed at $250

Output:

🔴 VERDICT: DROP

"That is an HP EliteBook 840 G5, equipped with an 8th Gen Intel Core i5 processor.

Market Value: Comparable units are selling right now for $120 to $170.

Your Price: $250

The Verdict: You would be significantly overpaying. For $250, you could easily 
find a much newer G7 or G8 model with significantly better performance. 

My Advice: Walk away from this one."

• Flask - Web framework
• Google Gemini 3.0 Pro - Multimodal AI for video analysis
• LangChain - Agent orchestration
• Selenium - Web scraping (eBay price data)
• ElevenLabs - Text-to-speech for voice responses
• Beautiful Soup - HTML parsing

• Vanilla JavaScript - No framework overhead
• Modern CSS - Glassmorphism design
• Font Awesome - Icons
• Custom Fonts - Clash Display & General Sans

• Custom eBay product search tool
• Real-time price comparison engine
• Verdict decision algorithm

• Python 3.9+
• Chrome/Chromium browser
• API Keys for:
  • Google AI (Gemini)
  • ElevenLabs

git clone https://github.com/optimusbuilder/thriftsmart-ai.git
cd thriftsmart-ai

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

pip install -r requirements.txt

Create a .env file in the root directory:

GOOGLE_API_KEY=your_google_api_key_here
ELEVENLABS_API_KEY=your_elevenlabs_api_key_here

macOS:

brew install chromedriver

Linux:

sudo apt-get install chromium-chromedriver

Windows: Download from ChromeDriver and add to PATH.

python app.py

Open your browser to: http://localhost:5001

thriftsmart-ai/
├── app.py                 # Flask backend & AI agent
├── system_prompt.py       # AI system instructions
├── requirements.txt       # Python dependencies
├── .env                   # Environment variables (create this)
├── static/
│   └── index.html        # Frontend interface
├── uploads/              # Temporary video storage (auto-created)
└── README.md             # You are here!

1. Upload a Video - Record or select a product video (MP4 format supported)
2. Enter Store Price - Type in the asking price
3. Click "Check Value" - Wait 30-60 seconds for analysis
4. Get Your Verdict - See if it's a COP or DROP + hear AI explanation

1. User uploads video + price
         ↓
2. Flask receives and saves video
         ↓
3. Video encoded to base64
         ↓
4. Gemini AI analyzes video content
         ↓
5. AI agent identifies product
         ↓
6. Selenium scrapes eBay (headless)
         ↓
7. AI compares prices & makes verdict
         ↓
8. ElevenLabs generates voice response
         ↓
9. Response sent to frontend
         ↓
10. User sees verdict + hears analysis

1. Product Identification (Gemini Vision)
   └─> Extract: Brand, Model, Specs
   
2. Market Research (eBay Scraping)
   └─> Find: Comparable listings, Price range
   
3. Price Analysis (LangChain Agent)
   └─> Compare: User price vs Market price
   
4. Verdict Generation
   └─> If savings > 30%: COP ✅
   └─> If savings < 10%: DROP ❌
   
5. Voice Synthesis (ElevenLabs)
   └─> Convert text response to speech

• Supports MP4
• Max file size: 500MB
• Uses Google Gemini's multimodal capabilities
• Identifies products from visual cues

• Headless Chrome browser (invisible)
• Searches eBay for comparable items
• Extracts product names and prices
• Returns top 10 most relevant results

• Analyzes market data vs asking price
• Considers product condition and age
• Provides detailed reasoning
• Clear COP/DROP recommendation

• Natural-sounding AI voice (ElevenLabs)
• Auto-plays analysis results
• Manual playback controls
• Visual waveform animation

• Glassmorphism design
• Smooth animations
• Drag & drop file upload
• Loading states with spinner
• Responsive layout
• Proper error handling

In app.py, modify:

audio = elevenlabs.text_to_speech.convert(
    text=message_content,
    voice_id="EiNlNiXeDU1pqqOPrYMO",  # Change this
    model_id="eleven_multilingual_v2",
)

Available voices: ElevenLabs Voice Library

In app.py, modify the determine_verdict() function:

cop_keywords = ['good buy', 'good deal', 'cop', ...]
drop_keywords = ['drop', 'pass', 'overpriced', ...]

Edit system_prompt.py to change how the AI analyzes products.

# Update Chrome
brew upgrade --cask google-chrome

# Reinstall ChromeDriver
brew reinstall chromedriver

Make sure you're accessing via http://localhost:5001 (not 127.0.0.1)

• Check browser console for errors
• Verify ElevenLabs API key is valid
• Click play button if auto-play is blocked
• Check system volume

• Ensure file size < 500MB
• Use supported formats only
• Check uploads/ folder permissions

• Video Analysis: ~10-15 seconds
• eBay Scraping: ~5-10 seconds
• Voice Generation: ~3-5 seconds
• Total Processing: ~30-60 seconds

Processing time depends on video size, internet speed, and API response times.

• Support for multiple marketplaces (Facebook Marketplace, Craigslist, etc.)
• Historical price tracking
• Product condition assessment
• Comparison with new product prices
• Save analysis history
• Mobile app version
• Chrome extension
• Batch analysis for multiple items
• Community price database
• AR product scanning

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

This project is licensed under the MIT License - see the LICENSE file for details.

Oluwaferanmi Oyelude

• GitHub: @optimusbuilder
• Email: [email protected]

• Google AI for Gemini API
• ElevenLabs for voice synthesis
• Anthropic Claude for development assistance
• The open-source community

Built with ❤️

⭐ Star this repo if you find it useful!

Made by Oluwaferanmi Oyelude | © 2026