Use Cases
Thoth for Researchers and Engineers
Research discussions often contain unpublished results, proprietary methods, and NDA-protected collaboration details. A recording of that conversation is as sensitive as the data itself. It should never leave your machine.
Why cloud recorders don't work here
- Pre-publication results become third-party data. Audio uploaded to a cloud recorder is held by a vendor subject to their own retention policies, security posture, and legal obligations. A data breach or law enforcement request reaches data you uploaded, not just data you intended to share.
- NDA-protected collaboration: cloud AI may count as disclosure. Sending a transcript to a cloud AI provider for summarization can constitute a disclosure under the terms of an NDA, depending on how the agreement defines third-party access. Most cloud recorders do not provide the contractual guarantees needed to rule this out.
- EU researchers face GDPR obligations for US-processed meeting data. When meeting audio or transcripts are processed on US servers, even for internal discussions, GDPR data transfer requirements apply. Those transfers require a legal basis, documentation, and in some cases prior supervisory authority review.
How Thoth fits
- Audio, transcript, and AI summaries never leave your Mac. No third party has access at any point. The only copy of your meeting data is on hardware you control.
- No data retention policy to review, no DPA to sign. On-device processing eliminates the compliance surface. There is no data transfer to document, no vendor agreement to audit, and no breach surface beyond your own machine.
- Works fully offline. Lab discussions, field work, and travel never require an internet connection. Transcription, speaker diarization, and AI summarization all run locally.
- Optional BYOK. If you prefer a more capable cloud model for summaries, the request goes directly from your Mac to the AI provider. Thoth never sees the transcript.