Privacy
Last updated 28 October 2026.
In short
- No audio is ever stored, by us or our speech-to-text provider.
- Transcripts are deleted after 30 days. Write-ups and Memory stay until you delete them.
- We never sell your data, never use it for advertising, and never use it to train AI.
- Only you can see your data. Export or delete all of it, any time.
Who we are
Sotto is run by [Company legal name], [street address], Philadelphia, Pennsylvania, USA ("we"). Questions about privacy: privacy@sottolens.com. We answer within 30 days, usually much sooner.
What we collect
- Account: your email, a scrambled (hashed) password, and your time zone.
- Glasses: a random device key, and when it was last used.
- Audio: streamed from your glasses' microphone, through the Even app on your phone, straight to our speech-to-text provider, only while you are in a session you started. It never reaches our servers, and nobody stores it.
- Text: the transcript of those sessions, and what Sotto wrote from it: notes on the lens, write-ups of meetings and conversations, to-dos, quizzes and your answers, study notes.
- Memory: who was in each conversation, what it was about, and a short summary of each person and project, written from your conversations. You can read, change, merge or delete any of it on your Memory page, or by telling Sotto.
- Notes and files you add: what you type, dictate or upload on the Memory page. Dictation uses your browser's own speech recognition, run by your browser's maker (in Chrome, Google; in Safari, Apple), before the text reaches us. From a file we keep the text, not the file: the text is read out of it (a photo is read by an AI model, which describes it and copies its words), turned into a note, and the file itself is not stored. Sotto doesn't recognise faces in images.
- Classes: what you type about your classes, and the pre-reads we write from it.
- About you: what you tell us about your work and how you want help. It goes into what the AI is told, so it can help you.
- Calendar: if you add a private calendar link, the link itself (stored encrypted, never shown back in full) and the events we read from it: titles, times, who is invited, descriptions.
- Plans: what you write on the Today page about your day and your meetings.
- Reminders: the time you chose and your browser's push address. The notification itself is a fixed reminder line with nothing about you in it.
- Your own keys: if you use them, stored encrypted and never shown again, only their last four characters.
- Search index: a numeric fingerprint of each write-up (an "embedding") so Ask can find it. It is deleted with the write-up.
- Messaging: if you connect Telegram, your Telegram chat ID and username, and what you and Sotto say there, which is kept with your Ask chats. Voice notes are turned into text and the audio is not kept; files follow the same rule as files you add here. Telegram also processes these messages under its own policies. Disconnect any time in Settings.
- Consent: when you agreed at sign-up to tell people, and whether you chose to skip the lens's "Told them?" question.
- Usage: how many minutes you listened, how many words the AI wrote and how many web searches ran, to run your plan.
- Website visits: the page and a random id your browser keeps. No cookies and no advertising trackers.
Why
Only to run the service for you: transcribe your sessions, write what appears on your lens, keep your notes and memory, and bill your plan. We do not sell your data and we do not use it for advertising.
We don't train AI on your data. Ever. We do not use your transcripts, notes, Memory, or anything said by the people around you to train or improve any AI model, ours or anyone else's. There is no "de-identified" or "anonymised" exception. Every provider we use is on terms that forbid training on what we send. If we ever want to change this, we will ask you first, it will stay off unless you switch it on, and it will never cover anything recorded before you said yes. On your own keys, your own account settings with those providers also apply; see "Using your own keys" in the Terms.
People around you
While a session is on, Sotto hears everyone nearby, not just you. Here is exactly what happens to what they say:
- Their voices: sent from your phone straight to our speech-to-text provider and turned into text. Neither we nor that provider keep the sound.
- Their words: kept in your transcript for 30 days, then deleted automatically.
- What Sotto writes from them: your write-ups of meetings and conversations, your to-dos, and the short notes about people and projects on your Memory page (for example, "Priya: leads the pricing project; prefers email") stay until you delete them or your account. Because they are written from what was said, they can contain things other people said.
- Telling voices apart: within one session, the transcript labels voices "Speaker 1", "Speaker 2" so it reads properly. Sotto does not recognise anyone by voice, does not build a voice profile of anyone, and nothing about a voice carries from one session to the next.
- Only you can see any of it. We never sell it, never use it for advertising, and never use it to train AI.
- If someone asks you to delete what Sotto has about them, you can forget them on your Memory page in a tap. If they come to us, see People near a Sotto user.
Who processes it
Our providers are listed on Subprocessors, with what each one receives.
How long we keep it
| What | How long | You can |
|---|---|---|
| Audio | Never stored, by us or our speech-to-text provider | — |
| Transcripts | 30 days, then deleted automatically | Forget the last minutes; delete a conversation |
| Nudges and AI answers | With their session's transcript | Same |
| Write-ups, to-dos, study notes, quizzes | Until you delete them | Delete; export |
| Memory (people, projects) | Until you delete them | Edit, merge, forget, on the Memory page or by telling Sotto |
| Notes you add, and the text read from files | Until you delete them. The files themselves are not kept | Delete |
| Calendar link and events | Until you disconnect the calendar | Disconnect |
| Consent record | With your account | Delete your account |
| Account and billing | Account: until deleted. Invoices: as tax law requires (kept by Stripe) | Delete your account |
| Server logs | Never contain what was said. Kept by our host for [period to confirm] | — |
| Backups | Encrypted, rolled over every [period to confirm], so deletions reach backups within that time | — |
| Website visit records | 13 months | — |
Your choices
- Export everything as one file, in Settings.
- Forget the last minutes of a conversation, from its page.
- Forget a person or project in Memory.
- Delete your account, which deletes everything at once.
- Stop a session or pause listening from the glasses at any time.
Schools
Sotto is a personal tool you sign up for yourself. It isn't provided by your school, and your school doesn't control your Sotto data.
Children
Sotto is only for people 18 and over. Don't use it to record conversations with children unless a parent or guardian agrees, and don't use it in schools for under-18s. If we learn an account belongs to someone under 18, we'll close it and delete its data.
Where Sotto works
Sotto is available in the United States, and your data is stored in the United States.
Permissions on the glasses
Microphone: only during a session you start, to transcribe it. Network: to reach our service.
If there's a breach
If we learn that someone got into Sotto data without permission, we'll email the people affected without undue delay, aiming for within 72 hours of confirming it. We'll say what happened, what data was involved, what we've done, and what you can do. We'll also notify regulators where the law requires.
Changes
We'll email you at least 30 days before a change to this policy that matters.