Scope
Voice memo capture, transcription, summaries, and a clean mobile workflow.
Projects
Flagship shipped product
Sonora
An AI-powered voice memo app built to turn capture into usable notes quickly. Concept to App Store in 18 days.
Recording 00:42
Summary
18 days from concept to App Store
12 transcription languages supported
97% Clean Architecture compliance target
Problem
Voice notes are easy to create and too easy to lose.
Sonora was built around a simple constraint: capture should stay fast, but the memo should become useful without a second workflow. The app combines recording, transcription, and summary-oriented processing in one focused iOS surface.
Architecture
Small app, production boundaries.
I treated Sonora as a real product from the start: clear domain boundaries, persistence that survives lifecycle changes, and AI services isolated behind interfaces instead of scattered through views.
- Swift 6 + SwiftUI interface layer
- Use-case driven Clean Architecture
- OpenAI Whisper transcription path
- Persistent memo storage and background processing
- Live Activities for active recording state
Timeline
Architecture
SwiftUI feature slices, Clean Architecture boundaries, and service abstractions for AI work.
Polish
Live Activities, multilingual transcription, persistence, and release hardening.
Ship
Submitted and launched on the App Store as a complete iOS product.
Outcome
The proof was shipping.
Sonora became the flagship example for the site because it compresses the story: taste, architecture, AI tooling, mobile craft, and release discipline in one public artifact.
Open the App Store listing