ArtMatch

ArtMatch is a React Native mobile application that revolutionizes art discovery through AI-powered recommendations, a Tinder-style swipe interface, AR preview, and a conversational art chatbot.

Comment

Inspiration

The global art market is worth over $65 billion, yet it remains one of the most inaccessible industries for everyday collectors. First-time buyers face overwhelming choices, lack knowledge about what matches their taste, and struggle to visualize how art will look in their spaces. Meanwhile, Singapore’s vibrant gallery scene remains largely undiscovered by locals and tourists alike.

We asked ourselves:

What if discovering art could be as intuitive as swiping through a dating app?
What if AI could understand your aesthetic preferences and guide you to your perfect artwork?

ArtMatch was born from the vision of democratizing art collecting by combining a familiar swipe interface with the power of AI-driven personalization.

What It Does

ArtMatch transforms art discovery into an engaging, personalized experience:

  • Smart Onboarding
    A 60-second style quiz presents artwork pairs to learn your aesthetic preferences and budget range, building a personalized taste profile.

  • Swipe to Discover
    Browse a curated feed of artworks with Tinder-style interactions. Each piece includes AI-generated explanations of why it matches your taste.

  • AR Preview
    Visualize exactly how an artwork looks on your wall before committing. Drag to reposition, adjust viewing distance, and see accurate dimensions in your space.

  • AI Art Advisor
    Chat naturally with our Claude-powered assistant. Ask questions like:
    “Find me colorful abstract pieces under $2000 near Orchard”
    and receive personalized recommendations with clear explanations.

  • Gallery Discovery
    Explore Singapore’s art galleries through an interactive map, with personalized recommendations, directions, and contact details.

  • Collection Management
    Track favorites, monitor your collection’s total value, and analyze your evolving style preferences over time.

How We Built It

Frontend

  • React Native application built with Expo for cross-platform development
  • React Native Reanimated for smooth swipe gestures and animations
  • Expo Camera for AR preview functionality
  • React Navigation with a bottom tab structure
  • Zustand for lightweight state management

Backend

  • FastAPI server handling API requests
  • Supabase (PostgreSQL) for data storage
  • Recommendation engine using scikit-learn, implementing a hybrid collaborative + content-based filtering algorithm that improves with each swipe

AI Integration

  • Claude AI powers:
    • Conversational chat interface
    • Personalized explanations for artwork recommendations
  • The AI understands user preferences, budget, and interaction history to deliver context-aware suggestions

Database

  • Supabase-hosted PostgreSQL database containing:
    • 12 real Singapore galleries with accurate locations
    • 55+ artworks across 10 different art styles
  • Tracks user preferences, swipe history, and collections to continuously improve recommendations

Challenges We Ran Into

  • AR Implementation
    Without native AR frameworks, we built a camera overlay system that simulates artwork placement with distance-based scaling for realistic perception.

  • Recommendation Cold Start
    New users lack swipe history. We solved this with an engaging onboarding quiz that enables personalized recommendations from the first swipe.

  • Real-Time Personalization
    Balancing recommendation quality with response speed required optimizing our ML pipeline using efficient feature extraction and caching.

  • Data Scarcity
    Comprehensive datasets for physical artworks in Singapore galleries were limited. We created realistic mock data designed for easy replacement through future gallery partnerships.

Accomplishments We’re Proud Of

  • Built a fully functional end-to-end application within a hackathon timeframe
  • Created a smooth, intuitive swipe interface with consistent 60fps performance
  • Integrated Claude AI for genuinely helpful, context-aware art recommendations
  • Designed an AR preview system without relying on complex AR SDKs
  • Developed a recommendation engine that learns and improves with every interaction
  • Mapped 12 real Singapore galleries, creating meaningful local value

What We Learned

  • AI UX Design
    Designing AI to enhance—not replace—human decision-making, keeping users in control.

  • Mobile Animation Performance
    Mastering React Native Reanimated to deliver complex gesture-driven animations at 60fps.

  • Recommendation Systems
    Hands-on experience with hybrid recommendation algorithms combining content-based and collaborative filtering.

  • Rapid Prototyping
    Making fast architectural decisions and prioritizing features that demonstrate core value under tight deadlines.

  • Full-Stack Integration
    Strengthening skills across React Native, FastAPI, databases, and real-time AI interactions.

What’s Next for ArtMatch

  • Gallery Partnerships
    Replace mock data with real gallery inventories, enabling direct purchases and gallery visits.

  • Secondary Marketplace
    Allow collectors to list and sell artworks, increasing liquidity and fostering a collector community.

  • Enhanced AR
    Implement ARKit/ARCore for accurate wall detection, lighting simulation, shadows, and reflections.

  • Social Features
    Enable users to follow collectors, share collections, and discover art through community curation.

  • Investment Insights
    Integrate market data to show price trends, artist trajectories, and potential investment value.

  • Provenance Verification
    Explore partnerships or blockchain-based solutions to verify artwork authenticity and ownership history.

  • Regional Expansion
    Extend beyond Singapore to galleries across Southeast Asia, becoming the region’s go-to art discovery platform.

Built With

Share this project:

Updates