Privacy Policy

Effective date: 2026-02-21

MoltSpeak (“we”, “our”, “the app”) is a speaking-under-pressure training app. This policy explains how we collect, use, and share information when you use the app.

Microphone & Audio Recording

The app requests microphone permission (android.permission.RECORD_AUDIO) to record your speaking responses during training reps. Audio is recorded on your device using Expo’s expo-av library.

Transcription & Analysis

When you complete a speaking rep, the app uploads that rep’s audio clip to our backend service at https://moltspeak-worker.moltspeak.workers.dev for transcription. Our backend forwards the audio to OpenAI’s transcription API and returns the transcript text to the app.

The transcript text may also be sent to our backend for analysis (for example, feedback and scoring). Our backend forwards that text to the OpenAI Responses API to generate analysis results.

Reflection audio is recorded in the app, but in the current version it is not uploaded by the app’s code path.

Data Storage & Retention

• Server-side: In our current backend code, we do not store audio files, transcripts, or analytics in a database or object storage. Audio and text are processed and forwarded to OpenAI to provide transcription/analysis.
• On-device: The app stores a generated device identifier and usage-related flags in local storage (AsyncStorage). Session transcripts and scores are kept in memory during the session and are not persisted by the app’s code.
• Audio files on device: The app does not explicitly delete recorded audio files; they may remain in temporary/cache storage managed by the operating system until cleaned up.

Third-Party Services

We use third-party services to provide transcription and analysis:

• Cloudflare Workers (our backend endpoint) receives audio uploads and analysis requests.
• OpenAI receives audio files for transcription and transcript text for analysis.

Logging

Our backend may log limited diagnostic information. In the current backend code, logs may include small snippets of model output, which can contain transcript-derived text fragments.

Your Choices

• You can deny microphone permission, but the app’s core recording features will not work.

Contact

Email: privacy@moltspeak.com