TasteMate AI — Project Story

Inspiration 💡

Choosing a movie to watch, a restaurant to visit, or a hotel to stay in can be surprisingly difficult when there are so many options available. Traditional recommendation platforms don't always make the discovery process feel personal or simple.

I wanted to build something that makes these decisions easier through a conversational experience. This inspired me to create TasteMate AI, a personalized recommendation platform that helps people discover movies, restaurants, and hotels based on their preferences, interests, mood, and budget.

My goal was to create an experience where users don't have to search endlessly. Instead, they can answer a few simple questions and discover options that better match what they're looking for.

What It Does 🚀

TasteMate AI provides three recommendation categories:

• Movies: Discover movies based on mood, genre, and viewing preferences.
• Restaurants: Explore dining options based on food preferences, budget, and dining needs.
• Hotels: Discover accommodation options based on destination, budget, and preferred stay experience.

The platform uses a guided chatbot to collect user preferences and presents recommendations through visual cards with relevant details and explanations.

Qloo Taste Intelligence is intended to help power the recommendation experience by using relationships between tastes, interests, and entities.

How I Built It 🛠️

I built TasteMate AI using a separate frontend and backend architecture.

Frontend

• React and TypeScript
• Vite
• Tailwind CSS

Backend

• Node.js
• Express.js
• TypeScript
• REST APIs

The frontend provides an interactive chatbot experience, while the backend manages conversations, processes preferences, and handles recommendation requests.

I designed the architecture to support Qloo integration on the backend so that API credentials remain private and are never exposed to the browser. I also developed a local mock-data mode to test the user experience and recommendation flow during development.

Challenges I Faced 🧩

One of the biggest challenges was designing a conversational experience that works across three different categories. Movies, restaurants, and hotels require different questions and preferences, so I needed to make each flow relevant without making the experience complicated.

Another challenge was separating the frontend and backend while keeping communication between them consistent. I worked on API integration, environment configuration, error handling, and organizing the project for development and deployment.

Access to the hackathon's Qloo API credentials was also a challenge during development. I prepared the integration structure and used clearly identified mock data for local testing while waiting for access to live API functionality.

What I Learned 📚

Building TasteMate AI helped me improve my skills in full-stack development, API integration, environment configuration, and designing conversational interfaces.

I also learned the importance of separating application components, handling external API failures, protecting credentials, and distinguishing test data from real API results.

Most importantly, I learned that a useful recommendation system needs more than a list of results. It needs to understand what the user is looking for and make the discovery process easier.

What's Next 🔮

My next steps are to complete and validate the live Qloo integration, improve recommendation relevance, and refine the experience across all three categories.

I also want to improve recommendation explanations and explore additional ways to make discovering movies, restaurants, and hotels more intuitive.

TasteMate AI is my attempt to turn everyday choices into a simpler, more personalized discovery experience.

Built With

• gemni
• node.js
• qloo
• react
• typescript