Launching today
Yogurt
Free, private, open-source alternative to Granola
15 followers
Free, private, open-source alternative to Granola
15 followers
Yogurt takes AI meeting notes without joining your call as a bot. It records Zoom, Google Meet and Teams in the background, shows a live transcript, and writes a summary with next steps when the call ends. Ask questions across all your meetings, a folder or one call, even mid-meeting. Replay the full audio, share notes with Claude or ChatGPT via MCP, and recordings stop when the call does. Free, open source, no account. Everything stays on your computer. Works on Mac/Windows/Linux.
Free
Launch Team
Framer AI AgentsDesign and publish professional sites with AI
Promoted
Hi Product Hunt! I’m Christian, co-founder of Frigade. Eric and I built Yogurt in our sparetime because we wanted useful AI meeting notes without sending our conversations to the cloud or paying for another subscription.
This is now possible because small open-weight models that can run on most computers are now good enough for meeting notes. You don’t need a frontier model to summarize a call or answer questions about it.
Yogurt is a free, open-source, MIT-licensed meeting notes app for Mac, Windows, and Linux. It records in the background with Zoom, Google Meet, and Teams, and never joins your call as a bot (but don't forget to ask people permission before you record them!). Everything runs locally: your recordings, transcripts, and summaries stay on your computer. There isn't an account or a paid tier available. Truly, we just felt taking AI notes shouldn't cost you money.
A few details we care about:
Ask questions across all your meetings, a folder, or a single meeting (even during call)
Echo detection when you’re on speakers, plus recordings that automatically end when your call does.
Calendar integration (Mac only, Linux/Windows coming soon)
A local MCP server so you can choose to share notes directly with Claude, ChatGPT, or other AI apps, without a paid cloud middleman.
Hear the full audio transcription
Under the hood, transcription uses Apple’s on-device speech on supported Macs, with whisper.cpp available everywhere. Summaries and chat use Gemma through llama.cpp, with a model picked and downloaded automatically for your hardware.
We’d love feedback on the first-run experience and the quality of your first meeting summary. Was setup clear, and did the summary capture what mattered?
Thanks for taking a look. Eric and I would love to hear how you use it!