Hoverfly - Tiny on-device AI models for Kotlin Multiplatform apps

Hi Product Hunt! 👋 I'm Raju, an Android developer and the maker of Hoverfly. Every model here started as a problem I hit in my own apps. I wanted simple AI features like emoji suggestions, smart replies or hiding a phone number before a message is saved. The usual answer was a cloud API: send the user's text to a server, pay per call, wait for the network, and worry about privacy. So I built the opposite: small models that live inside the app. 🧠 8 models, one job each: Moji (emoji), Beacon (language detection), Comma (punctuation), Comeback (smart replies), Emotion, Gatekeeper (toxicity), Hideout (personal info hiding) and Chalk (doodles) 📱 One Kotlin Multiplatform library each: Android, iOS, desktop and web ⚡ Milliseconds per call, 2–9 MB each, fully offline 🔒 Nothing leaves the device: no server, no API key, no account 🇮🇳 Built for real chat: Hinglish and Indian languages work, not just English How it changed along the way: it started as Android-only libraries. I rewrote all 8 in pure Kotlin Multiplatform so the same model gives identical results on every platform, and I test that against the reference model on each one. It's free up to 10,000 monthly active devices, so you can ship it today. If you need a model for your own task, I also build custom ones. I'd love your feedback: which model would you use first, and what should I build next? 🙏