Inspiration
Group decision-making is hard — we’ve all been stuck in endless WhatsApp threads trying to pick a movie, plan a trip, or choose where to eat. Everyone has different tastes, and too often, the “plan” never materializes.
We were inspired to create a system that doesn’t just recommend what’s popular — but what’s meaningful for a group based on their collective preferences. Our goal was to build a cultural mediator that helps groups find common ground and plan unforgettable experiences using AI.
What it does
CultureCircle is a multi-user, AI-powered experience planner that understands group preferences across domains like:
-Music
Movies & TV
Travel destinations
-Food & cuisine
-Fashion
- Books & podcasts
It uses Qloo’s Taste AI™ and Gemini LLM to:
-Analyze everyone’s preferences and compute a Harmony Score.
-Recommend movies, destinations, playlists, fashion and more — personalized for the group.
How we built it
CultureCircle is a full-stack web app built with the following stack:
Frontend:
-React.js with Tailwind CSS for a dynamic UI.
Backend:
-Node.js + Express.js for REST APIs.
-Integration with:
-Qloo API
-Gemini 1.5 Pro via Vertex AI for LLM processing
Database & Auth:
-Firebase Authentication for login/signup (email & Google).
-Firestore for storing:
-User preferences (e.g., favorite artists, movies)
-Group composition
-Shared insights and session data
Challenges we ran into
-Mapping user-friendly input to valid Qloo entity IDs was tricky.
-Understanding how to tune Qloo’s signal and filter parameters per entity type.
-Balancing token limits and latency when querying Gemini with rich group data.
-Designing the Harmony Score formula to be interpretable and fair.
-Coordinating multi-user state in Firebase with live updates.
Accomplishments that we're proud of
-Built a functional end-to-end system combining two powerful AI platforms.
-Designed an intuitive UI/UX for group input and experience browsing.
-Created a novel Harmony Score metric with LLM-based explanations.
-Successfully integrated dynamic group suggestions in real time.
-Made AI feel like a cultural companion — not just a chatbot.
What we learned
-Hands-on experience with Qloo’s cultural graph and entity modeling.
-Prompt engineering with QLOO API to reason about group behavior.
-Firestore rules and schema design for real-time multi-user collaboration.
-Optimizing performance of AI workflows inside frontend/backend pipelines.
What's next for CultureCircle
-Integrate with Spotify, Netflix, and OpenTable for action-ready suggestions.
-Expand to support mood-based planning and live polls.
-Add deep analytics for taste evolution and time-based recommendations.
-Launch a mobile app version with notifications and geolocation features.
-Build a group AI personality that evolves with your circle.
Built With
- axios
- express.js
- firebase
- node.js
- react
- tailwind
- vite
Log in or sign up for Devpost to join the conversation.