Show HN: AI-generated blackjack game built in about 3 hours (github.com)

A small project—an AI-generated, real-time Blackjack game—was created in roughly three hours and published as a production-ready Flask app. The game was scaffolded by an AI ("My‑AI") and iteratively refined by the author, delivering a playable site with a three‑level AI opponent, persistent player balances, and real-time multiplayer synchronization. It’s notable as a concrete example of how modern code-generation tools can rapidly produce not just toy demos but an architecture-minded app that addresses scaling, persistence, and UX responsiveness. Technically, the app uses Flask + Flask‑SocketIO (WebSockets) for instant client/server state sync, Redis as a session store so GameSession objects survive across Gunicorn workers, and SQLAlchemy (SQLite/Postgres) to persist balances. Eventlet handles asynchronous background tasks (AI and dealer turns) to keep the UI non‑blocking. The repo includes a clear project layout, .env handling, and deployment notes (e.g., set REDIS_URL to 127.0.0.1 on Windows). For AI/ML practitioners, the project demonstrates practical engineering choices—stateful session sharing with Redis, background task patterns, and how simple game AI tiers can be integrated into a scalable web service—making it a useful reference for rapid prototyping of interactive, stateful AI applications.