Privacy Policy

Last updated: April 18, 2026

Overview

Waffler is a free, open-source desktop app with no servers and no backend. It runs entirely on your machine. This policy explains what data we collect (virtually nothing) and how the app works.

Audio & Transcription Data

Waffler does not collect, store, or process your audio or transcription data. When you use Waffler, your audio is sent directly from your device to your chosen AI provider (Groq, Cerebras, or OpenAI) using your own API key. After transcription, the text is cleaned up by the same provider — again under your own key. We never see any of it. Your transcription history is saved locally on your device only, in ~/.waffler-hosted/history.json on macOS / Linux, or %USERPROFILE%\.waffler-hosted\history.json on Windows.

API Keys

Your API keys (Groq, Cerebras, or OpenAI) are stored locally on your device only — in ~/.waffler-hosted/.env on macOS / Linux, or %USERPROFILE%\.waffler-hosted\.env on Windows. They are never transmitted to us. All API calls go directly from the Waffler app on your machine to the provider's servers — we are not in the middle.

Account Data

No account is required to download or use Waffler. Waffler is free and open source — there are no purchases, subscriptions, or license keys. We do not collect or store any personal data.

Website Analytics

This website uses Cloudflare Web Analytics, which is a privacy-respecting analytics tool that does not use cookies or collect personal data. It provides aggregate traffic data only.

Data We Don't Collect

  • Audio recordings
  • Transcription text
  • Your API key
  • Usage patterns or analytics from the desktop app
  • Personal information of any kind
  • Cookies on this website

How Each Provider Handles Your Data

Waffler connects directly to your chosen API provider for transcription. Since you bring your own API key, how your audio data is handled depends on which provider you use:

Groq

Groq does not use API input data to train their models. Audio sent via the API is processed for transcription and not retained beyond what's needed to complete the request. Servers are based in the United States. This makes Groq the most privacy-friendly default option.

Groq Privacy Policy →

Cerebras

Cerebras does not use API input data to train their models. Inference runs on Cerebras's own data centres in Sunnyvale, California — the model itself (Qwen-3 235B) is open-source from Alibaba but the weights run on Cerebras hardware, so requests never touch Alibaba infrastructure. Subject to US data protection laws.

Cerebras Privacy Policy →

OpenAI

OpenAI's data retention and training policies vary by plan and API tier. As of May 2026, data sent through the API is retained for up to 30 days for abuse monitoring and is not used for model training by default. However, policies may change — check their current terms.

OpenAI Privacy Policy →

Bottom line: All three providers are US-based, store data on US infrastructure, and explicitly state they don't train on API input. Groq is the privacy-friendliest default; Cerebras adds speed; OpenAI provides redundancy. Waffler itself never sees, stores, or has access to your audio — it goes straight from your device to your chosen provider.

Contact

If you have questions about this privacy policy, get in touch at hello@wafflerai.com.