Ghummi - Cultural Travel AI

Inspiration

I recently traveled to NYC and it was horrible - I followed popular internet guides but nothing matched my personal taste. Every "must-see" spot felt generic and crowded. I realized travel recommendations are built for the masses, not individuals. What if we could match people to places based on their unique cultural DNA instead of tourist trends?

What it does

Ghummi is an AI-powered travel companion that creates hyper-personalized itineraries using cultural intelligence. Users describe their aesthetic preferences in natural language (favorite films, music, books, design styles), and Ghummi finds venues worldwide that share their cultural DNA. Instead of recommending popular tourist spots, it discovers hidden gems that align with your personal taste - like finding minimalist cafes for Wes Anderson fans or vinyl record shops for indie music lovers.

How we built it

Ghummi combines three powerful APIs in a sophisticated pipeline:

  • OpenAI GPT-4o analyzes user preferences to extract cultural themes and aesthetic DNA
  • Qloo's Cultural AI leverages 3.7+ billion cultural entities to find venues with matching aesthetic sensibilities
  • Google Places API provides real photos, ratings, and venue details for authenticity

Built with React/TypeScript frontend, Express.js backend, and Tailwind CSS for responsive design. The app features customizable trip lengths (1-7 days), venue counts per day, interactive itinerary editing, PDF export, and comprehensive sharing capabilities.

Challenges we ran into

  • API Integration Complexity: Orchestrating three different APIs with varying response formats and rate limits required careful error handling and fallback strategies
  • Cultural Matching Algorithm: Translating abstract aesthetic preferences into concrete venue recommendations while maintaining accuracy across 200+ global cities
  • Real-time Performance: Balancing comprehensive cultural analysis with user experience - achieving <30 second response times while processing complex taste profiles
  • Global Location Support: Building intelligent location normalization to handle spelling variations, common mistakes, and local naming conventions

Accomplishments that we're proud of

  • Successfully integrated Qloo's Cultural AI for the first time, unlocking true cultural intelligence in travel recommendations
  • Built a fully functional app with global coverage (200+ cities) in just days
  • Created an intuitive UX that makes complex AI-powered cultural analysis feel effortless
  • Achieved 100% consistent branding and professional-grade user experience
  • Implemented comprehensive sharing features including PDF export with clickable Google Maps links

What we learned

  • Cultural Intelligence is Transformative: Moving beyond popularity-based recommendations to cultural DNA matching creates genuinely personal travel experiences
  • API-First Architecture: Combining specialized APIs (OpenAI + Qloo + Google Places) creates more powerful results than any single service
  • User Experience Complexity: Making sophisticated AI analysis feel simple requires careful attention to loading states, error handling, and progressive disclosure
  • Global Scale Challenges: Supporting worldwide travel requires robust location handling, cultural sensitivity, and performance optimization

What's next for Ghummi

  • Collaborative Itineraries: Multi-user trip planning with preference merging for group travel
  • Real-time Updates: Live venue information, hours, and availability integration
  • Mobile App: Native iOS/Android apps with offline itinerary access and location-based notifications
  • Community Features: User reviews, photo sharing, and cultural taste-based social connections
  • Advanced Personalization: Learning from trip feedback to improve future recommendations
  • Business Model: Premium features, travel booking partnerships, and venue discovery services for tourism boards

Ghummi transforms travel from following crowds to discovering culture - creating truly personal adventures that match who you are, not where everyone else goes.

Built With

Share this project:

Updates