CulturePilot AI — Find Your Next Favorite Thing
Inspiration
People discover movies, music, food, and travel experiences through their interests and cultural preferences. However, finding something new that genuinely matches a person's taste can be difficult.
I built CulturePilot AI to make cultural discovery more personalized and interesting. The idea was to create an AI-powered agent that helps people explore new experiences based on their interests, using Qloo's cultural intelligence.
What It Does
CulturePilot AI helps users discover recommendations across multiple categories:
- Movies: Explore films that match your interests.
- Music: Discover music and artists.
- Food: Find food experiences related to your tastes.
- Travel: Explore places and travel-related experiences.
- All Experiences: Discover recommendations across categories.
Users can select a category and enter their interests. The application then uses a multi-agent workflow to research and generate relevant recommendations.
How We Built It
CulturePilot AI follows a structured agent workflow:
- Planner: Organizes the user's request and determines the recommendation task.
- Researcher: Retrieves relevant cultural information using the Qloo Insights API.
- Recommender: Uses the retrieved information and user interests to produce recommendations.
- Reviewer: Reviews the proposed recommendations before they are presented.
- Executor: Prepares the approved results for the application.
The project integrates Qloo's API through its documented Insights endpoint. The frontend is deployed on Vercel, and the backend is deployed on Render.
Challenges We Faced
One of the main challenges was connecting the frontend to the backend and making sure the complete workflow operated reliably. We also had to work with Qloo's API structure, including the correct endpoint, authentication header, and entity types.
Another challenge was improving recommendation relevance and variety. For example, a food discovery experience should aim to offer different kinds of suggestions rather than repeatedly recommending similar options. Testing each category helped us identify areas for improvement.
What We Learned
Building CulturePilot AI helped us understand how to integrate a real external API into an AI application, coordinate multiple agent roles, handle backend and frontend communication, and test a complete workflow.
We also learned that a useful recommendation system needs more than generating suggestions: it needs relevant cultural data, thoughtful filtering, and a review step.
What's Next
We want to improve recommendation diversity, make location-aware discovery more useful, and continue refining the relevance of results across movies, music, food, and travel.
CulturePilot AI brings AI agents and cultural intelligence together to help people discover their next favorite thing.## Inspiration
Built With
- ai-agents
- artificial-intelligence
- cultural-intelligence
- fastapi
- javascript
- personalization
- python
- qloo
- react
- recommendation-system
- rest-api
- vercel
Log in or sign up for Devpost to join the conversation.