AI · Privacy and processing
How EchoPod processes audio and AI requests
EchoPod uses local storage for your library and cloud services for features that need transcription, translation, account management, crash reporting, or AI-generated explanations. A cloud feature sends only the material required to fulfill that request.
Data flow by feature
| Feature | Data sent | Service used by the current code | Purpose |
|---|---|---|---|
| Account | Email, login token, account metadata | Supabase | Authentication, profile, progress, deletion |
| Transcription | Audio or a retrievable audio URL | Cloudflare Workers AI or Groq, depending on the selected path | Create timed transcript text |
| Translation | Transcript text and language codes | Cloudflare, Google translation integration, SiliconFlow, or OpenRouter depending on configuration | Translate transcript segments |
| AI explanation/chat | The selected transcript context and your prompt | Configured AI model provider through the EchoPod worker | Explain grammar, meaning, and context |
| Diagnostics | Crash and performance metadata | Firebase | Find crashes and improve stability |
Storage and retention
- Imported originals and converted media are stored in the app's private documents directory until the item is deleted.
- Temporary server audio uploads use the EchoPod Cloudflare storage path and can be deleted through the app's audio deletion request.
- Translation caches and transcripts may be stored locally to avoid repeating work.
- Provider-side retention is controlled by the applicable provider and account configuration; EchoPod does not claim a universal provider retention period.
- An account deletion request enters a grace period. The published privacy policy currently states 30 days; signing in during that period can cancel the request.
Your controls
- Delete an imported episode to remove its local media files.
- Use account deletion in the app to schedule removal of account-related server data.
- Contact echopod@clothpath.com for access, correction, export, or deletion questions.
Does EchoPod use my content to train its own model?
EchoPod does not operate a proprietary training program described in the current product code. Requests are processed by configured external services under their applicable terms. This statement does not override a provider's own data policy.
Why can the provider change?
The service includes fallback and selectable provider paths for availability and performance. The exact provider used can depend on feature and deployment configuration.