Blog
AI/Voice
March 12, 20267 min

AI Images and Inline Dictaphone: What's New in v1.3.2

v1.3.2 is the most ambitious I've released. Two major features and a delicate technical migration.

Integrated AI Images

The generated image is directly attached to the task. The user asks the voice assistant and the LLM crafts an optimized prompt by analyzing the context. Stored in privat.task_attachments with type ai_generated.

The inline dictaphone

No more full-screen overlay. The dictaphone integrates directly at the bottom of the screen. The key optimization: the transcribed text appears before the end of the TTS. The transcription displays immediately while the audio generates in parallel.

Skip TTS option added: some users don't need the voice response. Double gain: faster response time and reduced costs.

The gesture handler migration

Golden rule: in any GestureDetector, use components from react-native-gesture-handler, never those from standard React Native. The bug was vicious: functional on iOS but not Android, or in debug but not in release. I went through every file to correct the imports.

Performance

No unnecessary re-renders, animation via useNativeDriver, aggressive cleanup with 30s timeout for AudioPlayerService. v1.3.2 marks a turning point: the app is starting to have that fluidity that distinguishes it from a simple to-do.