AI-powered game recommendations based on how much time you have and how you're feeling.
Most recommendation engines ask "what do you like?" PlayNxt asks two better questions: How much time do you have? and What are you in the mood for? The result is a recommendation that actually fits your current moment, not just your history.
Problem. "What should I play tonight?" is real decision fatigue. Most game-discovery tools sort by popularity or genre. Neither matches how people actually decide, which is by mood and by how much time they have.
Who it's for. Gamers picking what to play in a limited window.
Key decisions and tradeoffs.
- Recommend on mood and time, not genre. The input model matches the real decision instead of making people browse taxonomies.
- Curated catalog over raw size. I validated a curated catalog rather than chasing a bigger, noisier dataset. Better recs beat more titles.
- Invested in CI/CD early. GitHub Actions from the start, so one person could ship to two stores without breaking things.
What I learned. A different input beats a different algorithm. Asking "how much time, what mood" reframed the whole product around the question the user is actually asking.
PlayNxt/
├── api-service/ # FastAPI recommendation engine (Python)
├── mobile-app/ # React Native + Expo (iOS & Android)
└── web-admin/ # React admin dashboard
CI/CD via GitHub Actions to cloud deployment.
- Mood and time-aware recommendations: not just "what you like" but "what fits right now"
- Cross-platform: iOS and Android via Expo
- FastAPI backend: lightweight, fast recommendation engine
- Admin dashboard: manage content and monitor usage
- CI/CD pipelines: automated builds and deployments via GitHub Actions
# API service
cd api-service
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env
uvicorn app.main:app --reload
# Mobile app
cd mobile-app
npm install
npx expo startBuilt by Watchlight Interactive