Inspiration: Planning the perfect vacation is overwhelming - endless options, constantly changing prices, and the fear of missing out on great deals. We were inspired by the challenge of creating an intelligent system that could understand individual travel preferences and continuously monitor for personalized opportunities. The idea of multiple AI agents working together like a specialized travel team excited us - each agent focusing on what they do best while coordinating seamlessly.
What it does: VacationGenius is a multi-agent AI system that learns your travel preferences and finds personalized vacation deals in real-time. Users simply describe their dream trip in natural language, and our intelligent agents work together to:
- Scrape and analyze current travel deals from TripAdvisor
- Learn from your preferences and past trips
- Match deals to your specific interests and budget
- Send personalized alerts when perfect opportunities arise
- Continuously refine recommendations based on your feedback
How we built it We architected a sophisticated multi-agent system using modern technologies:
- Frontend (Next.js): Clean, intuitive interface where users input their travel preferences
- Backend API (Express.js): Central orchestration hub that coordinates all agents
- Scraping Agent (Node.js + Apify): Continuously monitors TripAdvisor for new deals and pricing changes
- Personalization Agent (Node.js): Learns user preferences and calculates match scores
- Database (Supabase/PostgreSQL): Stores user profiles, trip requests, and scraped deal data
- Real-time Communication (Redpanda): Enables agents to communicate and coordinate activities The system uses Prisma ORM for database management and implements JWT authentication for secure user sessions.
Challenges we ran into
- Database Schema Evolution: Initially struggled with complex multi-table relationships, eventually simplified to a clean 2-table approach (users and trips) that better matched our hackathon timeline.
- Agent Coordination: Orchestrating multiple independent agents while maintaining data consistency was challenging. We solved this by implementing a centralized backend API that all agents communicate through.
- Real-time Data Synchronization: Ensuring scraped data stays current while maintaining performance required careful optimization of our polling and caching strategies.
- Environment Configuration: Managing API keys and database connections across multiple services required robust environment variable handling and secure configuration management.
Accomplishments that we're proud of
- Successfully built a working multi-agent system that demonstrates intelligent coordination Created a clean, scalable architecture that can easily accommodate additional agents
- Implemented real-time data scraping and personalization algorithms Achieved seamless integration between frontend, backend, and database layers
- Built a system that actually learns and improves over time based on user interactions What we learned
- Multi-agent systems require careful orchestration - each agent needs clear responsibilities and communication protocols
- Database design should prioritize simplicity and performance over complex relationships, especially in hackathon settings
- Real-time data processing requires robust error handling and fallback mechanisms
- Environment management is crucial when working with multiple external APIs and services
- User experience should drive technical decisions - sometimes the simplest solution is the most effective
What's next for VacationGenius
- Enhanced Personalization: Implement machine learning models to better understand user preferences and predict ideal travel timing
- Expanded Data Sources: Integrate with more travel platforms beyond TripAdvisor for comprehensive deal coverage
- Mobile App: Develop native mobile applications for iOS and Android
- Advanced Analytics: Add detailed travel insights and trend analysis for users
- Social Features: Allow users to share trips and get recommendations from friends
- AI Chat Interface: Implement conversational AI for more natural trip planning interactions
- Enterprise Features: Develop business travel optimization tools for corporate users The foundation we've built provides a solid platform for scaling into a comprehensive travel intelligence system that could revolutionize how people discover and book their perfect vacations.
Built With
- apify
- cloud
- database
- docker
- express.js
- framework)
- github
- javascript
- next.js
- node.js
- postgresql
- prisma
- react
- redpanda
- service)
- sql
- supabase
- typescript
Log in or sign up for Devpost to join the conversation.