Lisna vs cloud-based tools
What you get when transcription, structuring, and storage all run on your Mac.
| Feature | Lisna | Otter | Fireflies | Notion AI |
|---|---|---|---|---|
| On-device transcription | ||||
| Notes stay on device | ||||
| No data sent to LLM provider | ||||
| Real-time captions | ||||
| Markdown / Obsidian export | partial | |||
| Works offline | ||||
| Lecture-aware structuring | partial | partial | ||
| Free tier | ||||
| Price | $0 (alpha) / $? Pro | $8.33/mo | $10/mo | $10/mo |
Why we built Lisna differently
Cloud transcription is fast to build but loud about your data. Audio is uploaded to a vendor, transcribed on their GPUs, structured by their LLM, and stored on their servers. For students and researchers handling lectures, drafts, and unpublished ideas, that flow is wrong.
Lisna inverts it. Whisper runs on your Mac's Neural Engine. Llama 3.2 runs in your Mac's RAM. Notes write to your filesystem in Markdown — sync them with Obsidian or iCloud or no one if you prefer.
This means Lisna is slower on first launch (model downloads). It means we can't ship feature parity with cloud-only tools on day one. We think the trade is worth it.