AI-Powered Shopping Intelligence for Amazon & Daraz
Features • Tech Stack • Screenshots • Architecture • Ethics
TrustCart helps online shoppers make confident, informed decisions by analyzing product trustworthiness, tracking prices, and detecting fake reviews using AI.
- 🤖 Detects fake reviews using advanced AI pattern recognition
- 📊 Tracks price history and alerts you to the best deals
- 🎯 Provides trust scores (0-100%) for products and sellers
- 💡 Recommends actions: BUY, WAIT, or AVOID with clear reasoning
- 🔔 Sends smart alerts when tracked products hit target prices
- Fake review detection with confidence scores
- Trust ratings based on review patterns, seller history, and data quality
- Clear BUY/WAIT/AVOID recommendations with explanations
- Decision breakdowns showing key signals and warnings
- Real-time price tracking across Amazon and Daraz
- Historical price trends with visual charts
- Smart price drop alerts
- AI-powered "best time to buy" recommendations
- Smart watchlist with priority organization
- Personalized insights based on shopping behavior
- Comparison history and saved analyses
- First-time user onboarding tips (dismissible)
- AI confidence levels disclosed for all recommendations
- Minimum data requirements enforced (won't analyze <3 reviews)
- Admin monitoring for AI quality and abuse detection
- Input sanitization and rate limiting
- React 18 + TypeScript - Modern component architecture
- Vite - Fast development and optimized builds
- Tailwind CSS + shadcn/ui - Consistent, accessible design system
- Zustand/Context - Lightweight state management
- React Router v6 - Client-side routing
- Axios - HTTP client with auth interceptors
- Node.js + Express + TypeScript - Type-safe REST API
- MongoDB + Mongoose - Document database with schema validation
- OpenAI GPT-4 - AI analysis and recommendations
- Firebase Admin SDK - Authentication verification
- Node-Cron - Scheduled price tracking jobs
- Winston - Structured logging with log levels
- Firebase Auth - Google OAuth + Email/Password
- MongoDB Atlas - Cloud database
- OpenAI API - AI model access
- Rate Limiting - Per-route protection
- Input Sanitization - Security middleware
Landing page with AI-powered shopping intelligence
Feature highlights and benefits
Get started section
Personalized dashboard with watchlist and insights
Price tracking statistics and alerts
AI-powered shopping behavior insights
Recent activity and price changes
Detailed product view with AI trust score and analysis
Price history chart and recommendations
Customize currency, store preferences, and AI settings
Configure email alerts and notification preferences
Price drop alerts and admin notifications
Admin control panel with system metrics
User management and system health monitoring
System maintenance and AI quality monitoring
backend/src/
├── modules/ # Feature modules (auth, ai, products, etc.)
├── models/ # MongoDB schemas
├── services/ # Business logic
├── controllers/ # HTTP handlers
├── middleware/ # Auth, validation, rate limiting
├── utils/ # Helpers (logger, sanitizer, etc.)
└── config/ # DB, OpenAI, Firebase setup
- Separation of Concerns: Controllers ≠ Business Logic
- Type Safety: Full TypeScript across frontend and backend
- Error Boundaries: React error boundaries prevent white screens
- Caching: In-memory caching for expensive AI/scraping calls (30min scrape, 24hr AI)
- Rate Limiting: Per-route limits (e.g., 10 AI requests/hour)
- Structured Logging: Winston with levels (info, warn, error)
TrustCart uses OpenAI GPT-4 to:
- Analyze review text for patterns (fake vs. genuine)
- Calculate trust scores based on multiple signals
- Generate BUY/WAIT/AVOID recommendations
- Explain decisions with key signals and confidence levels
- Minimum Data Requirement: Refuses analysis with <3 reviews
- No Medical/Financial Advice: System prompts prevent harmful claims
- Confidence Disclosure: All AI responses include 0-100% confidence
- Human Oversight: Admin panel monitors AI usage and quality
- Bias Mitigation: Regular audits of AI outputs
✅ Transparent: Show confidence levels and data quality
✅ Honest: Acknowledge limitations and uncertainty
✅ User-Controlled: Users can dismiss tips, delete data
✅ Privacy-First: Minimal data collection, no selling data
✅ No Dark Patterns: No tricks or manipulative design
- Caching Layer: 30min for scraping, 24hr for AI analysis
- Optimistic UI: Watchlist updates feel instant
- Error Recovery: Friendly fallback UI instead of crashes
- Loading States: Skeleton screens and subtle indicators
- Code Splitting: Lazy-loaded routes for faster initial load
- Structured logs with Winston (no console chaos)
- Admin audit panel (AI usage, rate limits, errors)
- Error boundaries with "Reload" and "Go Home" actions
- Firebase handles auth (Google OAuth, Email/Password)
- JWT tokens verify all API requests
- Role-based access (User, Admin)
- Passwords never stored (Firebase manages)
- HTTPS for data in transit
- Input sanitization (prevents injection)
- MongoDB IP whitelist
- Environment variables for secrets
- AI-powered fake review detection
- Price tracking with alerts
- User authentication and profiles
- Admin dashboard
- Onboarding system
- Caching layer
- Error boundaries
- Browser extension
This project is licensed under the MIT License. See LICENSE for details.
In an era of fake reviews, manipulative pricing, and information overload, TrustCart empowers shoppers with:
- AI-backed confidence - Don't guess, know the trust level
- Price intelligence - Buy at the right time
- Transparency - See how decisions are made
- Control - Your data, your rules
TrustCart isn't just a tool—it's your ally in the modern e-commerce landscape.
Built with ❤️ for smart shoppers everywhere