Cleft has transcribed your voice on-device since the first builds. The cloud only ever handled the summary step. A look at why we made that choice early, and the work to make on-device transcription fast and accurate.
CarPlay is the most constrained surface we have shipped: you cannot look at the screen, you only get templated controls, and you might lose signal at speed. Here is how those limits shaped the feature.
1.12 through 1.12.3 added almost no new buttons on purpose. We spent the quarter making the app you already use harder to break.
The hard part of a voice app is capturing clean audio on real devices that connect late, switch mid-sentence, and get interrupted by calls. Here is the work that went into making recording survive that.
People asked us for a black hole they could dump a thought into and trust it was captured and synced everywhere. Sync had drifted from that, so for 1.10.1 we rebuilt the sync layer.
Justin published Liquid Speech, an MIT-licensed Flutter package for using Apple's native SpeechAnalyzer API on iOS and macOS while still compiling cleanly on older OS versions.
During an early React Native build of Cleft, react-native-skia would not compile for Mac Catalyst. I took a stalled community fix, brought it current against main, and got macOS Catalyst support merged upstream in v2.3.0.