Inspiration
We wanted to make cultural discovery feel like an adventure rather than scrolling through endless recommendations. What if a favourite movie could lead you to a book, an artist, or a podcast you never knew you would love? And what if familiar cultural experiences could help us create meaningful moments with the people we care about? That's how Curio was born.
What it does
Curio is an AI-powered cultural discovery companion built around two experiences: Personal Discovery and Memory Lane.
Personal Discovery helps users explore personalised recommendations across movies, TV shows, books, podcasts, and artists. Users can describe what they are looking for through natural language, explore cultural connections, plan their next five discoveries, build a Taste Profile, and collect discoveries in their Cultural Passport.
Memory Lane helps caregivers create personalised reminiscence sessions for their loved ones using familiar cultural references, personal interests, and life experiences. Caregivers can generate conversation starters, record reactions, and use feedback to shape future sessions.
How we built it
We built Curio using Python, FastAPI, JavaScript, Qloo's Taste Intelligence API, and Google Gemini.
Qloo provides the cultural intelligence foundation, helping Curio resolve cultural entities and retrieve relevant recommendations. Gemini interprets user intent, helps organise discoveries into meaningful journeys, and generates personalised explanations and reminiscence sessions.
We orchestrated these capabilities into multi-step workflows, with validation and filtering to keep results grounded in relevant cultural data.
Challenges we ran into
Working with diverse cultural categories and inconsistent API metadata was challenging. We had to handle entity resolution issues, filter irrelevant results, prevent duplicate discoveries, and preserve user context across multi-step journeys.
Another challenge was keeping Gemini grounded in Qloo's retrieved results rather than allowing it to generate unsupported recommendations. For Memory Lane, we also had to consider how personal context and caregiver feedback could be used to create relevant, meaningful sessions.
Accomplishments that we're proud of
We brought two distinct experiences together under one cultural-intelligence foundation.
For Personal Discovery, we built an interactive experience featuring personalised recommendations, natural-language exploration, multi-step discovery, taste profiling, and Cultural Passports.
For Memory Lane, we developed a caregiver-focused experience that transforms familiar cultural references into personalised reminiscence sessions and conversation starters.
Most importantly, we turned our idea into a working, deployed application that people can explore and use.
What we learned
We learned that building useful agentic AI applications requires more than connecting an LLM to an API. Reliable data retrieval, thoughtful orchestration, validation, state management, and a coherent user experience are equally important.
We also discovered that cultural intelligence has applications beyond entertainment and recommendations. Familiar cultural references can help create opportunities for connection, conversation, and shared experiences.
What's next for Curio
We aim to improve cross-category discovery and recommendation relevance, make cultural connections more explainable, and evolve the Cultural Passport into a richer record of each user's cultural journey.
For Memory Lane, we hope to expand personalisation, improve how caregiver feedback shapes future sessions, and make it easier to create meaningful reminiscence experiences for loved ones.
