I just released "Explain Yourself," a local multiplayer party game (Jackbox style) where players have to give excuses for absurd AI-generated scenarios. An AI Judge then ranks the answers, roasts the players, and determines a winner.
I built this because I wanted an AI-first party app game that was fun and made people use their brains. This is my first app, but I have spent months on it and it is pretty thoroughly thought out.
The Stack:
Frontend: SwiftUI (using NavigationStack and EnvironmentObjects for state).
Backend: Firebase (Firestore for real-time syncing, Cloud Functions for the game logic). AI: Gemini API (via Cloud Functions) for both generating scenarios and the "Judge" persona. The Technical Challenges:
Prompt Engineering: Tuning the "Judge" to be fun and a bit savage was a fun part. I also expected Gemini (which I ended up choosing for simplicity over other AIs) to not respond to many of the players answers if completely inappropriate, but it always does... so far.
The judge analyzes the answers and decided which was the most logical with a bit of creativity. The game scores 90% of one and 10% on the other to determine the ranking.
Monetization: It’s free to play with a daily limit/starter rounds. I hate ads, so I limited them strictly to "watch to earn" if you run out of credits, plus a standard IAP model for specific question packs. I wonder if starting out I should simply give out everything for free though, leaving only a couple of things like the custom categories as the locked IAP part. Would love some insight on that if you got any.
I’d love feedback on the latency of the app in general, the quality of the AI responses, and the IAP model most of all. Means the world. Thank you in advance.