Why Your Meeting Notes Should Never Leave Your Mac

I spent years building privacy tools. Products that protect millions of people’s data every single day. I’m pretty proud of that work.

So you’d think I’d know better than to invite a cloud meeting recorder into a call about our product roadmap. 😅

But I did. We all did. The convenience was too good. Join a Zoom, let the bot record, get a tidy summary ten minutes later. Magic.

Except here’s what’s actually happening: that bot is capturing every word. Your voice, your tone, the exact things your colleagues said about competitive strategy or that candidate you’re not sure about. And it’s shipping it all to a server you don’t control.

I didn’t think twice about it until I read the fine print on one of these services. They were using “de-identified” recordings to train their AI models. De-identified in quotes because… if the recording contains your Q3 roadmap discussion, it doesn’t matter if your name’s attached or not. The content is the identity.

This isn’t a theoretical risk

Think about what actually gets said in meetings.

You’re discussing which competitor you’re going to undercut on pricing. You’re giving candid feedback about a VP candidate. You’re on a call with lawyers about an acquisition that hasn’t been announced yet. Every one of these conversations is now sitting on someone else’s server.

And here’s the part that really got me: companies spend months getting SOC2 or GDPR compliant. They audit every database, every internal tool. Then an employee invites a cloud meeting bot into a call and punches a massive hole right through that compliance fence.

Nobody notices because it doesn’t feel like a data export. It feels like a productivity tool.

The trade-off that no longer exists

For years, there was a real reason we needed the cloud for this. Transcription was too heavy for a laptop. You needed big server clusters to turn speech into text.

That’s just not true anymore.

Apple Silicon Macs can handle high-quality transcription and AI summarization right on the device. No network calls. No servers. The audio stays in your machine’s memory and on your disk.

When I realized this, I thought: why is anyone still sending their meetings to the cloud?

So I built Oaken Notes

The idea was simple. A meeting recorder should be as private as a physical notebook. If you wouldn’t mail your handwritten meeting notes to a random startup for “processing,” you shouldn’t do it with your audio either.

Oaken Notes runs entirely on your Mac. Here’s what that actually means:

Transcription happens through WhisperKit: speech-to-text conversion that never leaves your system’s memory. Summarization uses Apple’s on-device language models. Your product roadmap doesn’t go anywhere.

We don’t use Sentry, Firebase, Mixpanel, or any analytics SDK. We don’t even require an account. You open the app and start recording. That’s it.

And because “we promise we’re private” isn’t good enough (I’ve been in this industry long enough to know that), we added a Privacy Pulse, a real-time indicator that shows exactly where your data is at any moment. You can see it while you work. No trust required.

The most secure server is the one that never gets your data

I love what AI meeting notes can do. Being able to actually focus on the conversation instead of typing is genuinely better. But we shouldn’t have to trade our professional confidentiality for that convenience.

The location of your AI matters. And the safest place for your meeting data is the device that’s already sitting in front of you.

If that sounds right to you, check out oakennotes.com.