I’ve been experimenting with Apple’s new on-device Foundation Models (iOS 18 / macOS Sequoia) and built a small tool around them: AppReview AI, a Mac + iPad app to analyze App Store reviews privately and offline.
The motivation was simple: as an indie developer, reading hundreds of reviews from competitor apps was slow, noisy, and hard to extract signal from. Existing tools rely on cloud processing, API keys, or external servers. I wanted something lightweight and private that used Apple’s new local AI instead.
What it does - Summarizes reviews using Apple’s on-device models - Extracts sentiment, recurring issues, bugs, and feature requests - Shows per-country ratings to detect market differences - Displays basic estimated downloads and revenue (SensorTower public data) - Syncs selected apps and analyses through iCloud
All AI processing stays on-device. No external servers, no accounts, no OpenAI key.
Why I built it I wrote a recent article on Apple Foundation Models and was surprised how far the local models can go with the right prompts. This project was a way to test how practical on-device analysis could be in a real use case for developers.
Free tier The app has a small free tier (1 app + 3 AI analyses) so anyone can try it without registration.
If you’re curious, here’s the link: https://apps.apple.com/lu/app/appreview-ai-review-analyzer/i...
I’d love feedback, criticism, or ideas for what to analyze next (keywords, rankings, crashes, changelogs, etc.). Happy to answer technical questions about the on-device AI integration as well.