Privacy
What happens to your lectures
The short version: recording, transcription and search happen on your own computer and your audio never leaves it — and unless you ask us to keep it, the recording is deleted once the transcript exists. With the AI features on, the transcript text is sent to a model run by someone else. Everything below is a claim about how the app is built, not an intention.
What never leaves your computer
- Your recorded audio, and any video you add
- The files you add — slides, PDFs, photos
- The transcript itself, which is produced on your machine
- Search, which runs against a database in the app's own folder
Speech is turned into text by a model running on your computer, not by a service. Text on pages and photos is read there too, by macOS's own recognizer. Nothing in this list is uploaded anywhere, whether the AI features are on or off.
What is sent, when the AI features are on
Some features need a model too large to run on a laptop. Those send the transcript text — the whole lecture, a block at a time — and text read from your slides and photos, in order to: fix misheard words, build each lecture's study memory, explain the moments you pinned, and write notes, flashcards, quizzes and the other study tools. On Pro, reading handwriting, charts, diagrams and equations also sends an image of the page being read.
It goes to My Prof's own server, which holds the model key so your computer never has to. That server keeps a usage record — which account, which model, how many tokens, when — and never the text or the image. From there, generated notes go to Google's Gemini API; if Google cannot take a request it goes to OpenRouter, which sends it to the same Google model. Everything else goes to AI models on OpenRouter.
While My Prof uses Google's free tier, Google may use what it receives to improve its products, and people at Google may read it; Google says it first separates it from our account. Do not pin or upload anything you would not want someone else to read.
Google, OpenRouter and the providers OpenRouter routes to handle what they receive under their own policies, which we do not control.
Other people's voices
A lecture recording contains other people. The transcript includes what anyone in the room said — your professor, and anyone who asked a question — and that text is part of what gets sent when the AI features are on.
You can decide this for yourself. You cannot decide it for them. Recording a class may also be governed by your instructor's permission, your institution's policy and the law where you are. Those rules are yours to follow; the app does not check them for you.
How long anything is kept
We do not keep a recording of anyone's voice. Audio is transcribed on your computer, and both files — the one you play back and the one the speech model reads — are deleted once transcription, indexing and pin matching have all succeeded. What remains is text. A lecture that fails or has not finished processing keeps its recording so it can be retried. You can turn on Keep recordings in the app's settings if you would rather hold on to them; then they stay in the app's folder until you delete them.
Transcripts, notes, pins and study material stay in the app's folder on your computer until you delete them. We hold no copy of any of it. The one thing our server holds is your account — the email address you signed in with, your plan, and that usage record.
Text already sent to a model provider is outside the app, under their retention policy rather than ours. Deleting a lecture on your machine does not reach back and delete it there.
Deleting your data
Deleting a class deletes its lectures, transcripts, pins, notes and study material with it. Because it is all files in one folder on your computer, deleting that folder removes every trace of the app's data from the machine. To close your account and remove it from our server, write to us and say so.
Turning the AI features off
They are on unless you turn them off, in the app's settings. When off, nothing is sent: everything on your computer keeps working, misheard words are still cleaned up offline, and pins are saved but not explained. The choice takes effect on the next request, including work already queued in the background.
What the website itself collects
This site has no analytics, no advertising and no third-party trackers. It loads its typeface from Google Fonts, which sees the request. Signing in sets one cookie, which holds your session and is read only by our server. Paying sends you to Stripe, whose own page handles your card — we never see it.