An intelligent conversational assistant that transforms online shopping with AI-powered product discovery, sentiment analysis, and personalized recommendations.
ShopSense isn't just another e-commerce chatbot. It's a sophisticated AI-powered shopping companion that combines:
- 🧠 Intelligent Intent Recognition - Understands complex shopping queries and classifies intent (product search, price comparison, reviews, general chat)
- 💬 Context-Aware Conversations - Maintains conversation history with persistent chat sessions stored in real-time
- 🎯 Semantic Product Search - MongoDB text indexing for lightning-fast, relevant product discovery
- 📊 Advanced Sentiment Analysis - Dual-mode sentiment processing using both Gemini AI and custom ML models for accurate review analysis
- 📧 Smart Email Generation - AI-crafted product recommendation emails with personalized content
- 🔊 Text-to-Speech Integration - Convert responses to natural speech for accessibility
- 🔐 Secure Authentication - Appwrite-powered auth with OAuth support and guest mode
- 🎨 Beautiful UI/UX - Modern, responsive design with dark mode and 40+ shadcn/ui components
Online shoppers face three critical challenges:
- Information Overload - Thousands of products, reviews, and specifications make decision-making exhausting
- Trust Deficit - Difficulty assessing genuine product quality from mixed reviews
- Time Consumption - Hours spent comparing prices, reading reviews, and researching alternatives
ShopSense solves these problems by:
- Conversational Product Discovery - Ask questions naturally: "Find me wireless headphones under $100 with good battery life"
- Instant Review Synthesis - AI-powered sentiment analysis summarizes hundreds of reviews in seconds
- Smart Comparisons - Compare products side-by-side with pricing insights and feature breakdowns
- Proactive Recommendations - Email yourself curated product suggestions for later review
- Natural language product queries with intent classification
- Context-aware responses based on conversation history
- Real-time streaming responses for instant feedback
- Guest mode for quick access, authenticated mode for personalized experience
- Hybrid sentiment engine combining Gemini AI and custom ML models
- Review summarization with positive/negative/neutral categorization
- Highlighted examples from actual customer reviews
- Confidence scoring for reliability
- MongoDB text search with relevance scoring
- Multi-field indexing (name, category, description)
- Price range filtering and sorting
- Availability status tracking
- AI-generated personalized product recommendation emails
- SendGrid integration for reliable delivery
- Customizable templates with product details and links
- One-click email sending from chat interface
- Text-to-speech conversion using Gemini TTS
- Screen reader compatible components
- Keyboard navigation support
- WCAG-compliant UI design
- Responsive design for all devices
- Dark/light theme toggle with system preference detection
- Smooth animations and transitions
- Real-time typing indicators
- Next.js 15.0.3 - React framework with App Router
- React 18.3.1 - UI library
- TypeScript 5.6.3 - Type-safe development
- Tailwind CSS 3.4.17 - Utility-first styling
- shadcn/ui - 40+ beautiful, accessible components
- Radix UI - Unstyled, accessible component primitives
- Next.js API Routes - Serverless functions
- Google Gemini AI - Chat completions, sentiment analysis, email generation, TTS
- Appwrite 16.0.2 - Authentication, realtime database, storage
- MongoDB 6.20.0 - Product database with text search indexing
- SendGrid - Transactional email delivery
- @google/generative-ai - Google Gemini integration
- @xenova/transformers 2.17.2 - Client-side ML models for sentiment analysis
- Custom Intent Classifier - Query categorization engine
- Sentiment Analyzer - Hybrid sentiment processing
- Docker - Containerization for consistent deployments
- Vercel - Production hosting with automatic deployments
- GitHub Actions - CI/CD pipeline for testing and validation
# Pull the pre-built image
docker pull himaenshuu/shopassist:v1
# Run the container
docker run -p 3000:3000 --env-file .env himaenshuu/shopassist:v1
# Access at http://localhost:3000Docker Image: 🐳 himaenshuu/shopassist:v1
- Node.js 18+ and npm
- MongoDB Atlas account (free tier works)
- Appwrite account (cloud or self-hosted)
- Google Gemini API key
- SendGrid API key (for email features)
- Connect GitHub repository to Vercel
- Add environment variables in Vercel dashboard
- Deploy automatically on push to main
Automated testing and validation on every pull request:
- ✅ ESLint - Code quality and style checks
- ✅ TypeScript - Type validation
- ✅ Build - Compilation verification
Protected main branch ensures only validated code reaches production.
User Query → Intent Classifier → Product Service → MongoDB
↓
Gemini AI
↓
Response Generation
↓
User Interface
Intent Classification
- Analyzes user queries to determine intent
- Routes to appropriate service (product search, review analysis, chat)
- Confidence scoring for accurate classification
Product Search Pipeline
- MongoDB text indexing for fast full-text search
- Relevance scoring and ranking
- Price filtering and availability checks
Sentiment Analysis Engine
- Gemini AI for semantic understanding
- Fallback to custom ML model for reliability
- Structured output parsing with confidence metrics
Email Generation
- Context-aware content creation
- Product detail integration
- SendGrid delivery with tracking
We welcome contributions!
- Fork the repository
- Create a feature branch
- Make your changes with tests
- Submit a pull request
This project is licensed under the MIT License - see the LICENSE file for details.
- Google Gemini AI - Powerful language models
- Appwrite - Backend-as-a-Service platform
- MongoDB - Scalable document database
- Vercel - Deployment and hosting
- shadcn/ui - Beautiful component library
- Open Source Community - Amazing tools and libraries
- Issues: GitHub Issues
- Discussions: GitHub Discussions
- Email: himanshusingh26.2.2004@gmail.com
Built with ❤️ using Next.js and Google Gemini AI
⭐ Star this repo if you find it helpful!