TasteWeave AI — Your Taste Is a Pattern

Inspiration

Our favorite music, movies, fashion, and lifestyle choices are not isolated preferences. They often reflect a deeper cultural identity.

Yet most AI recommendation systems treat these interests separately or generate suggestions from generic language-model knowledge. We wanted to explore a different approach: what if an AI agent could connect the things people love and use those connections to create recommendations that feel more personal?

That idea inspired TasteWeave AI, a culturally intelligent styling and lifestyle agent powered by Qloo's taste intelligence.

What it does

TasteWeave AI turns a user's cultural interests into a personalized taste profile and styling experience.

Users select favorite artists, musicians, films, fashion aesthetics, lifestyle interests, and colors. The application brings these preferences together into a visual Taste DNA profile, including an interpreted style archetype, a color palette, and personalized lifestyle recommendations.

The intended live workflow uses Qloo's cultural intelligence API to retrieve relevant entities and cultural signals. An agent workflow then uses those results as evidence when generating recommendations.

Our goal is to move beyond recommendations that simply sound plausible toward suggestions with traceable cultural grounding.

How we built it

We are building TasteWeave AI as a full-stack web application using Next.js, React, and TypeScript.

The application includes an interactive, multi-step onboarding experience, a Taste DNA dashboard, an editorial-inspired interface, and a structured recommendation workflow.

We are developing a provider architecture that separates the user experience from the cultural intelligence backend. This allows us to build and test the experience with clearly labeled demo data while preparing the integration with Qloo's official hackathon API.

The application is designed to keep API credentials on the server and distinguish demo results from live API responses.

What makes it different

TasteWeave is built around the idea that cultural context can make personalization more meaningful.

Rather than relying exclusively on a user's explicit fashion choices, we want to explore how interests across music, film, fashion, and lifestyle can inform a coherent personal aesthetic.

A key part of the experience is transparency: recommendations should explain their supporting signals, distinguish retrieved evidence from AI interpretation, and allow users to refine their preferences.

Challenges

The main challenge is connecting cultural intelligence to useful, explainable recommendations.

A model can produce an attractive style description without having any evidence that the recommendation fits the user's cultural interests. Our architecture separates cultural retrieval, recommendation generation, and explanation so these responsibilities can be evaluated independently.

Another challenge is building a polished consumer experience without obscuring the distinction between sample data and actual API results. We are addressing this with explicit demo-mode labeling and a server-side integration design.

What we learned

This project has reinforced that an effective AI agent requires more than a language model. It needs relevant external data, structured workflows, careful handling of evidence, and an interface that helps people understand and refine the results.

We are exploring how Qloo's cultural intelligence can help an AI application move from generic suggestions toward culturally informed personalization.

What's next

Our next milestones are to complete the live Qloo integration, evaluate recommendations across different taste profiles, improve evidence explanations, and deploy a publicly accessible application.

Our vision is simple: help people discover the connections between what they love and find new experiences that feel authentically their own.

Built with: Next.js, React, TypeScript, and Qloo's cultural intelligence API.

Built With

• agents
• ai
• artificial
• fashion
• intelligence
• learning
• machine
• next.js
• personalization
• qloo
• react
• recommendation
• systems
• typescript