Voice workflows should stay clear about what is recorded, stored, and sent.

Summary

golosAway is designed around user-started voice actions. Audio and text are processed only after you begin a recording or translation flow.

Data processed

Based on the current workspace, golosAway may process microphone audio, transcribed text, translated text, preset settings, enabled destinations, enabled languages, keyboard-language preference, preset description context, and a user-supplied API key.

API key storage

The app stores the user’s personal OpenAI API key in Keychain and its settings copy explicitly says the key remains there rather than in app history.

OpenAI processing

The README states that golosAway uses OpenAI for speech-to-text and translation, with no separate developer backend included in the project.

Temporary audio

The workspace documentation says cancelled recordings are deleted immediately and temporary audio is deleted after dictation or translation completes.

No transcription history

The current settings screen states that golosAway stores no transcription history.

Keyboard flow state

The keyboard flow uses shared app-group state so the app and keyboard extension can coordinate active recording status, active preset details, pending insertion, request IDs for stop-and-insert, and keyboard-language preference.

Conversation extensions

The current project also includes conversation-level extension flows such as the Messages extension, which means dictated or translated text can move through additional app-extension surfaces when you explicitly use them.

Preset context

Presets can include an optional description field, and that context may be sent together with the message to improve transcription or translation quality.

Extensions and handoff

The project includes a Share Extension and widget-related controls, which means data can move through app-extension flows when you explicitly use them.

Contact

For privacy-related questions, contact [email protected].

Last updated: August 9, 2026