Taste Mirror Cultural Intelligence App

🧠 Inspiration We were inspired by the idea that taste is more than preference — it's cultural identity. Music, fashion, food, and media choices reveal who we are, who we relate to, and how we view the world. Yet, most tech platforms treat taste as static or surface-level. We wanted to build something that goes deeper: an app that uses cultural signals to understand people, build connections, and uncover insights beyond demographics.

⚙️ What it does Taste Mirror analyzes a user’s taste profile (e.g., favorite music, brands, cuisines) and transforms it into a cultural intelligence report using AI and Qloo’s Taste Graph. The app:

Generates a Cultural Fingerprint — a multidimensional snapshot of identity

Maps users to taste clusters (e.g. "Neo-Retro Creatives", "Global Minimalists")

Shows cross-domain connections between interests (e.g., how your love of Kendrick Lamar connects to your fashion and travel preferences)

Provides cultural compatibility scores between individuals, groups, or brands

🏗️ How we built it Frontend: React (Chrome Extension + web app) with TailwindCSS

Backend: Node.js + Express API hosted on Vercel

LLM: OpenAI GPT-4 for identity narrative synthesis, taste analysis, and language understanding

Qloo API: Used to fetch related entities and audience graphs across multiple cultural domains (music, fashion, travel, food, etc.)

Storage: Lightweight use of localStorage for Chrome extension; potential expansion to Supabase for user history

🧱 Challenges we ran into LLM tuning: Getting GPT to consistently output structured cultural insights rather than just surface-level recommendations

Taste normalization: Mapping free-form user input (e.g. "I like Solange, oat milk, and linen") to Qloo-friendly entities

Cultural bias: Ensuring the model doesn’t over-index on Western or mainstream cultural clusters

UX balance: Presenting deep insight in a clean, non-overwhelming way for casual users

🏆 Accomplishments that we're proud of Successfully integrated Qloo’s API with an LLM layer to turn raw preferences into cultural intelligence

Created a usable Chrome extension that analyzes web content and overlays taste-based cultural interpretations

Developed a Cultural Fit Score between people or between a user and a brand/service

Built a system that doesn’t just say what people like, but why — and how it connects to others

📚 What we learned Taste is incredibly expressive, but hard to quantify — Qloo’s graph is powerful, but LLMs help contextualize it meaningfully

Personalization is about values as much as categories — it's not just "what you like" but "what that says about you"

LLMs are amazing at narrative — and surprisingly good at reverse-engineering cultural profiles

🚀 What's next for Taste Mirror: Cultural Intelligence App Real-time taste graph explorer: Interactive UI for seeing how interests link across domains

API for brands: Let businesses assess how well they align with audiences on a cultural level

Group identity mapping: Understand communities through collective taste signatures

Mobile app version with photo-to-profile feature (e.g., scan your closet or Spotify to build taste profile)

Open source Taste Profile Schema for ethical personalization and social research

Built With

• nextjs
• qloo