Show HN: Corgina – Open-source pregnancy tracker I built for my wife (iOS 26) (github.com)

Corgina is an open-source, privacy-first iOS pregnancy tracker that uses natural-language voice commands and GPT-4o to estimate nutrition, automatically log meals, hydration, vitamins and symptoms, and track PUQE nausea scores. The app runs on iOS 26 (TestFlight available), stores all health data locally (no analytics or cloud storage), and performs on-device speech recognition for privacy. Users can bring their own OpenAI API key (stored securely in the iOS Keychain) to enable AI nutrition estimation and voice parsing; the app also functions without a key for manual logging. Typical GPT-4o costs are modest (~$0.01 per food item, ~$0.005 per command; daily usage ~$0.10–0.30). From a technical perspective, Corgina is built in SwiftUI with Apple’s new Liquid Glass design, uses MVVM plus actor-based concurrency, and leverages the Speech framework for on-device transcription. Storage relies on UserDefaults and the local file system, notifications use UserNotifications, and the OpenAI integration is handled via an OpenAIManager/AsyncTaskManager flow. For developers, it’s practical to build and modify (requires Xcode 17+, macOS, Developer Mode on device) and demonstrates a hybrid architecture: privacy-preserving local UX combined with optional cloud-powered ML for richer estimations—useful as a reference pattern for future health-focused, privacy-first AI apps.