Inspiration
OmniTaste AI Engine was built to bridge the gap between AI conversation and true cultural preference discovery. By tapping into Qloo's vast cross-domain data, we wanted to create an agentic experience that understands human taste beyond simple keywords.
What it does
It is an interactive conversational taste agent that learns what you like and follows your taste into other domains (such as a film leading to a café, brand, city, or writer). All taste recommendations are strictly backed by the Qloo Insights API across four core pillars: Entertainment, Lifestyle & Travel, Fashion & Brands, and People & Culture.
How we built it
- Backend & Frontend: Integrated into a seamless full-stack deployment on Vercel using Python, FastAPI, and a React-based UI.
- AI & Data Integration: Powered by LLM integration for natural conversational flow and the Qloo API for deep cultural entity mapping across all entity types.
Challenges we ran into
Handling state persistence across serverless environments posed some session timeout challenges during extensive multi-turn testing. We implemented robust session management and clear navigation guides for users to reset sessions easily when needed.
Accomplishments that we're proud of
Successfully integrating multiple Qloo pillars into a smooth conversational AI agent and achieving a fully responsive, live deployment on Vercel.
What's next for OmniTaste AI Engine
Expanding recommendation granularity, enhancing multi-user profile comparisons, and refining state handling for even smoother extended conversations.
Log in or sign up for Devpost to join the conversation.