AI Wallpaper App—Learns your style—100% On-Device & Open
Vanderwaals uses on-device machine learning to understand your aesthetic preferences and automatically surfaces wallpapers you'll love—no endless scrolling required. 🤖 Neural network (100% offline) 🔒 Zero tracking, zero analytics 📚 3,000+ curated wallpapers from GitHub + Bing ⚡ Auto-change on unlock/hourly/daily 🎨 Material 3 with dynamic theming 🔓 Fully open source (AGPL-3.0) Two modes: Start fresh and let AI learn, or upload one favorite wallpaper for instant similar matches.
Vanderwaals is fully open-source with transparent code available on GitHub. Network requests are limited to wallpaper downloads and weekly catalog sync only—no hidden tracking, ads, or surprise data collection.
Vanderwaals provides 11,400+ curated wallpapers from GitHub collections (aesthetic, minimalist, anime, Linux ricing) and Bing Photography Archive (UHD professional photography), with weekly auto-updates adding new content.
All ML inference runs entirely on-device with zero analytics, no Firebase, no tracking SDKs, and no data collection. Your preferences never leave your phone, and the open-source code (AGPL-3.0) is publicly auditable.
Vanderwaals uses on-device MobileNetV4 ML to learn your visual preferences from likes/dislikes and automatically delivers matching wallpapers. Auto-change triggers (on unlock, hourly, or daily) keep your device fresh without manual effort.
The app uses Exponential Moving Average (EMA) algorithm to smoothly integrate feedback from likes, dislikes, and implicit signals (wallpaper duration). It balances your original taste with newly discovered preferences while occasionally suggesting new styles.