Before 1.0
The road to 1.0.
ParseMeet didn't start on-device. The first builds were a classic cloud pipeline: a Python backend, hosted speech-to-text, a hosted large-language-model API. It worked — and every meeting a user recorded would have had to travel to servers we operated.
In mid-2026, before ever shipping publicly, we deleted the entire backend. Not migrated — deleted: the servers, the upload path, the API keys, the accounts system. The app was rebuilt on Apple's private AI stack — speech recognition on-device, and Apple Foundation Models on your Mac, your iPhone, and Apple's Private Cloud Compute. Builds since then make zero calls to our servers, because there are none.
The privacy policy got shorter because the architecture did. 1.0 is the first public release, and on-device isn't a feature we added — it's the foundation everything else sits on.
Read the full story →