This is a submission for Weekend Challenge: Passion Edition
What We Built
Kyuutai is a fandom devotion platform for people who want to track, reflect on, and celebrate the things they love most. Instead of treating fandom like a casual list of favorites, it turns it into an interactive experience built around three passions: gaming, anime, and music.
The project is organized around four connected experiences:
- Devotion Log — a journal for recording what I watched, played, or listened to, with mood tags, voice dictation, and a timeline view.
- Weekly Recap — an AI-generated narrated summary of the last 7 days of fandom activity, with audio playback.
- Passion Card — a personalized passion radar that analyzes top games, anime, and artists into an archetype profile, recommendations, and a downloadable visual card.
- OST Detective — a soundtrack guessing game built around real game and anime clips, with AI-written questions and narrator reactions.
Team submission:
- Dylane(https://dev.to/yveane) — anime
- Ace — gaming
- Kanjo(https://dev.to/alchemycodes) — music
The goal was to make fandom feel personal, expressive, and a little dramatic — like a living diary for obsession, taste, and identity.
Demo
Live demo: https://kyuutai.vercel.app
Video demo:
Code
GitHub repository:
yvesdylane
/
kyuutai
Kyuutai is a passion journal with a twist: tell it what game, anime, or song owned your brain today - by voice or text - and it grows into a living timeline of your fandom.
.
How I Built It
Kyuutai is built with Next.js 16, React 19, TypeScript, Tailwind CSS 4, PostgreSQL, NextAuth, Google Gemini, and ElevenLabs.
I kept the app organized with a clean App Router structure and a route -> service -> repository pattern so the HTTP layer, business logic, and SQL stay separated.
Google Gemini powers the intelligence layer. It generates weekly recap scripts, analyzes favorite media into a structured passion profile, and writes the short question/reaction text used in OST Detective. Each feature uses a different prompt style: recap narration is warm and dramatic, passion profiles return structured JSON, and OST Detective uses short game-show-style lines.
ElevenLabs handles the voice layer. It turns recap scripts and passion blurbs into audio files saved under public/audio/, which makes the app feel more like a fandom experience than a standard dashboard.
The Devotion Log also uses browser speech recognition for hands-free note-taking, while OST Detective pulls real soundtrack clips from YouTube using yt-dlp and trims them with ffmpeg for short blind-test rounds.
On the database side, the app uses raw SQL with parameterized queries, UUIDs, and a schema designed around journal entries, recap history, passion cards, and game sessions.
Prize Categories
- Best Use of Google AI
Google Gemini is the intelligence layer behind multiple parts of Kyuutai, not just a single feature. It writes the weekly recap narration, transforms favorite media into a structured passion profile, and powers the short question and reaction text in OST Detective. I used different prompt styles for each experience: narrative generation for recap audio, strict JSON output for the Passion Card, and concise game-show style writing for the trivia game. That made Google AI central to both the app’s core functionality and its personality.
- Best Use of ElevenLabs
ElevenLabs brings Kyuutai’s emotional layer to life through voice. The app converts AI-generated recap scripts and passion blurbs into spoken audio, making the experience feel like a fandom show rather than just a dashboard. In practice, ElevenLabs is used to create weekly recap narration and the Passion Card voice profile, giving users something they can listen to, not just read. The audio output is part of what makes the app feel immersive and polished.
Top comments (0)