Blog
Notes on private, on-device AI. How local transcription actually works, and how to tell real privacy from a privacy policy.
-
Is Granola AI safe? What local-first actually covers
Granola AI is bot-free and does not store your meeting audio, which puts it ahead of most cloud notetakers. Your audio still goes to Deepgram, AssemblyAI, OpenAI and Anthropic to be processed. Here is the sourced version.
-
Two-party consent states, and what the lists get wrong
Most articles give you a clean list of two-party consent states. The clean list is the problem: counts range from 11 to 12, Michigan has a participant exception, and Oregon and Connecticut split by conversation type. Here is the honest version.
-
Is Otter.ai safe? What actually happens to your recordings
Otter holds SOC 2 Type 2 and encrypts everything, and it is also defending a federal wiretap case over how its notetaker joins meetings. Here is what Otter publishes, what the litigation alleges, and how to decide.
-
AI note takers that don't join your meeting: how they work
Bot-free notetakers capture audio from your device instead of dialling into the call. That solves the awkwardness, and it is a separate question from whether your audio still goes to the cloud. Here is how each approach actually works.
-
Local AI transcription on a Mac: what actually runs on your machine
How on-device transcription works on a Mac, what the models cost you in disk and battery, and how to check whether a tool that claims to be local really is.