💡 Inspiration

Standard Large Language Models generate tourist clichés. When asked: "Plan an intimate night out in Tokyo for someone who loves Thom Yorke and Wong Kar-wai," ChatGPT suggests mainstream tourist traps like Shibuya Crossing and basic sushi. It lacks any grasp of cross-domain aesthetic affinity — the subtle threads linking melancholic art-rock textures to brutalist concrete cocktail lounges, or 90s Hong Kong cinema lighting to subterranean natural wine bars.

🧠 What It Does

CulturAI eliminates cultural blindness by mathematically grounding Google Gemini 2.5 Flash in Qloo’s Taste AI™ Graph (575M+ cultural entities across Music, Film, Dining, Fashion, Books, and Travel). CulturAI computes the exact aesthetic intersection between musical textures, architectural dining spaces, and subterranean nightlife, producing hyper-personalized, uncompromised lifestyle realities in under 2 seconds.


⚡ The 8 Specialized Feature Engines

  1. 🍸 Night-Out & Date Curator (/night-out): Chronologically sequenced 4-part nocturnal journey (Twilight Sound -> Architectural Gastronomy -> Arthouse Cinema -> Secret Speakeasy) with integrated Spotify sound pairing and Google Maps routing.

  2. ✈️ Subculture Travel Concierge (/travel): Anti-tourist 3-day cultural residency mapping users into independent artistic districts (e.g., Daikanyama architecture, Shimokitazawa vinyl dens) across 45+ premier metropolises.

  3. 👥 DuoSync Bilateral Compromise (/duosync): Solves couple indecision ("What should we do tonight?") by computing the bilateral vector equilibrium between two distinct taste profiles with zero awkward compromise.

  4. 🧬 Cultural DNA Radar (/cultural-dna): 5-Dimensional lifestyle radar persona (Sonic Depth, Cinematic Aesthetic, Gastronomy Curiosity, Urban Subculture, Literary Resonance) generating shareable cyber-luxury identity archetypes.

  5. 🎬 Screenplay & Soundtrack Moodboard (/filmmaker): Empowers creative directors and auteurs by extracting lighting palettes, acoustic cues, and real-world filming locations directly from raw narrative concepts. Includes strict gibberish filtering.

  6. 🛍️ TasteMatch Cultural Commerce (/tastematch): Curates physical design artifacts, audiophile hardware, and archival fashion pieces (Issey Miyake, vintage tube amplifiers, rare vinyl) with live Qloo affinity percentages.

  7. 🤝 Group Hangout Consensus Solver (/hangout): Mathematical multi-user centroid solver for 3 to 6 friends with 100% strict veto compliance (e.g. Zero Loud EDM, No Tourist Traps).

  8. 📬 Weekly Cultural Radar Dispatch (/digest) & Saved Vault (/library): Monday morning bespoke cultural digest with tailored gallery openings, film retrospectives, and underground listening sessions, saved to a user-isolated Firestore vault.


🛡️ Dedicated Admin Suite & Real-Time API Diagnostics

To give judges 100% transparency into our architectural integrity and live Qloo production keys, CulturAI includes a full-featured Admin Suite:

  • 🎛️ Central Admin Control Center: https://culturai-app-nu.vercel.app/admin
    Inspect platform-wide metrics, live database collections, and user role management.
  • 🔑 Live Qloo API Diagnostics & Key Console: https://culturai-app-nu.vercel.app/admin/keys
    Judges can click to ping https://hackathon.api.qloo.com/search in real-time, inspect live HTTP 200 telemetry, or dynamically input their own Qloo Hackathon key to test instant validation.
  • 💎 Zero-Mock Verification Ribbon: Across all 8 engines, CulturAI displays authentic Qloo Entity UUIDs (e.g. Radiohead: 70CAE5BF-2F4C-445C-A3E5-4EDACFC3591C), seed origin tags, and mathematical affinity scores directly from Qloo servers.

🏛️ How We Built It

  • Framework: Next.js 16.4.2 (App Router & Turbopack), React 19.3.0, TypeScript 5.x.
  • Styling: TailwindCSS v4 with Cyber-Glassmorphism and deep void palette (#040409).
  • Cultural Intelligence: Direct live queries to Qloo Taste AI™ API (https://hackathon.api.qloo.com/search).
  • Autonomous Reasoning: Google Gemini 2.5 Flash via multi-agent orchestrator (agentOrchestrator.ts).
  • Cloud Persistence: Cloud Firestore (user isolation, dynamic API settings, itinerary vault) & Firebase Auth.
  • Audio & Mapping: Spotify Web API deep-links and custom SVG geodata coordinates.

🧗 Challenges We Overcame

  1. Handling Cold-Start Profiles: Designed the Progressive Taste Refinement Engine to detect skipped onboarding answers and dynamically prompt 1-click rich seeds without re-running onboarding.
  2. Harmonizing Cross-Domain Affinity Vectors: Blending orthogonal domains (e.g. ambient drone music with brutalist dining) into a coherent chronological night-out required rigorous multi-agent prompt conditioning.
  3. Sub-2.0s Dual-Engine Latency: Parallelized Qloo API queries and Gemini 2.5 Flash synthesis to achieve end-to-end multi-agent execution in under 2 seconds.

🏅 Accomplishments That We're Proud Of

  • 100% verified zero-mock production deployment connecting directly to Qloo's live Hackathon API.
  • 8 fully functioning autonomous engines delivered (far exceeding hackathon mandates).
  • 1-Click Judge evaluation access with zero setup friction.
  • Dedicated Admin & API Diagnostics console for instant judge verification.
  • Strict TypeScript verification with zero compilation errors (tsc --noEmit clean).
  • MIT Open Source repository published on GitHub.

⚡ Testing Instructions (For Judges)

  1. Instant 1-Click Evaluation: Visit https://culturai-app-nu.vercel.app and click "1-Click Direct to Dashboard" on the login screen. (Demo credentials if needed: Email: [email protected] | Password: judgeDemo2026!)
  2. Test Live Admin & API Diagnostics:
  3. Test Autonomous Engines: Navigate to /night-out, /travel, /duosync, or /hangout to test real-time cultural curation grounded in verified Qloo graph coordinates.

Built With

Share this project:

Updates

Submission history