Cathedra — Privacy Policy

Last updated 17 September 2026 · corrected, and the connector added, 29 September 2026 · the connector's calendar changes, 30 September 2026 · a shared lecture added to a library, a course shared as one link, and the connector's timetable changes, 2 October 2026 · the connector's cancelled and make-up classes, and its lecture corrections, 3 October 2026 · two versions of a screen compared, under usage data, 5 October 2026

Who holds it

Cathedra is made and run by Remzi Atasagun Poyrazoğlu, who is the data controller for everything described here. Anything on this page — a question, a request to see or delete what we hold, a complaint — goes to support@cathedra.study and is answered by a person.

What the app stores, and where

Everything Cathedra creates lives in the app's own storage area on your device:

• Audio recordings of lectures you choose to record, as files on the device.
• Transcripts produced from those recordings.
• Your timetable — terms, courses, class times, rooms, exams and deadlines.

Audio stays on the device that recorded it. A term of lecture audio is several gigabytes, and the app is built so it never goes anywhere — not to us, not to iCloud. Transcripts and your timetable sync to your own private iCloud database so your library follows your Apple Account to your other devices; Apple gives developers no access to a user's private database, so we cannot read any of it.

The AI features, if you use them

These are the only parts of Cathedra that send anything to another company, so each one is described in full. All but one of them need a subscription, and all but one wait to be asked; the exception is the same in both cases, it is free for everyone, it is the first one listed below, and it is the reason this paragraph no longer says "only when you ask". Cathedra AI is on unless you turn it off — this policy, which you accept when you sign in, names the services, and Settings turns it off in the app and on the website alike, at which point nothing here sends anything, including that one, and every other part of Cathedra works exactly as before.

• Fixing misheard terms is the one that runs by itself. A speech recogniser gets ordinary words right and technical terms wrong — "binary search" comes back as "final research" — and everything made from a transcript afterwards inherits the mistake. So when a lecture has finished being written down, your iPhone works out which phrases sound like a term from that course and sends only those phrases, with a few words around each, to be judged. The lecture itself is not sent, nothing is stored, and the answer is a list of yes and no. What it changed is kept on each line, so the app can put every word back.
• Finding announcements — exam dates, assignments, room changes — runs on Apple's on-device model. Nothing is sent anywhere, and it works with no connection.
• Asking a question about a lecture sends the text of that lecture's transcript to DeepSeek, an AI service outside your country, so it can answer you. Asking a question about a whole course sends the transcripts of that course's lectures, together with their titles, their dates and the course name. Your audio recording is never sent.
• Describing a photograph — a slide, a whiteboard — sends that photograph, and reading a syllabus sends the document or screenshot you chose. Only what you picked, and only when you ask.
• Speaking a question instead of typing it sends those few seconds of your voice to Groq, a speech-to-text service outside your country, so the words can be understood in any language. The text comes back; the audio is not kept.
• Making an audio summary sends the text of that lecture's transcript to DeepSeek, to write a short conversation about the lecture, and sends that conversation to Google, a speech service outside your country, to be spoken. Your audio recording is never sent. If you are signed in, the spoken summary is kept with your account so it plays on the website too, and it is deleted with the account; without an account it is kept only on your iPhone.
• Requests pass through a small service we operate on Cloudflare. It holds our API key, checks with Apple that your subscription is active, and counts how many questions the subscription has asked this month. To do that it keeps, for a few hours, the subscription's original transaction identifier and whether it was active, and the month's count — nothing else. It stores no transcript, photograph or document, and writes no content to any log. The one thing it keeps is an audio summary made for a signed-in account, described above.
• Your subscription is handled by Apple. We never see your name, email or payment details.

Your account, if you make one

You can use every part of the app without an account. An account exists for Cathedra on the web, and you can walk past the offer when the app first asks and sign in later from Settings, or never.

• An account is your email address and nothing else — no name, no password. You sign in by typing a six-digit code we email you; the code works for ten minutes.
• We keep the address, when the account was made and last used, and a record of each signed-in device (which kind, when it signed in, when it was last seen), on a database we operate on Cloudflare in Western Europe. The sign-in code and the session are stored only as hashes.
• If you sign in on your iPhone, your courses, your lectures' titles and dates, their transcripts, and your questions and answers are copied to your account on our server (Cloudflare, Western Europe), so the website can show them and Cathedra AI can answer from them there. Recordings are not copied. Signing out stops the copying; deleting the account deletes the copy.
• An audio summary you make while signed in is kept with the account in the same place, so the website and your other devices can play it. Deleting the account deletes it.
• What you agreed to is kept with the account: the version of the terms and of this policy you accepted, your Cathedra AI and usage data choices, and when you made each. That is why the app and the website never ask twice — an answer given on one is already known to the other.
• The only email we send is the sign-in code. Nothing else, and never to anyone else.
• Signing out ends that device's session. Deleting the account — in the app under Settings › Account, or on the website — removes the address and every session; it asks for a fresh code first. Your lectures, transcripts and timetable are not part of the account and stay where they are.

A lecture you upload or record on the website, if you do

The website can transcribe an audio file you upload to it — a lecture recorded on another device, say — and can record a lecture itself, on a laptop or an Android phone. Those are the only kinds of lecture audio that reach our server, and only because you sent them:

• The file is stored with your account on our server (Cloudflare, Western Europe) and sent to Groq, a speech-to-text service outside your country, to be transcribed. The transcript comes back to your account; Groq does not keep the audio.
• It stays with the lecture so a transcription that failed can be tried again. Deleting the lecture deletes the file, and so does deleting your account.
• It never goes to your iPhone, and a recording made on your iPhone never comes here.
• Recording on the website asks your browser for the microphone, and only while you record. The recording is sent a few seconds at a time, as it is made, to Groq to be transcribed, and its words come back to the page. That audio is not kept — not on our server and not by Groq — only the transcript is saved to your account. Pieces not yet sent stay in your own browser until they are, so a closed tab loses nothing, and are removed from it when the lecture is finished.

Your own AI app, if you connect one

With an account you can let an AI app of your choice — ChatGPT or Claude, say — read your library, and change your calendar and your timetable when you ask it to, through Cathedra's connector, after you sign in and approve it on our page.

• The app then receives what it asks for with its tools: your courses, your timetable and calendar, your lectures' transcripts and your documents' text. Without a subscription it receives only the last 7 days of lectures and documents.
• When you ask it to, it can add, move or remove the exams, deadlines and events in your calendar; for an app connected since 2 October 2026, whose approval page said so, add a course to your timetable, add, move or remove its weekly classes, and call one class off or add a make-up class; and for an app connected since 3 October 2026, correct a lecture you recorded — its title, the day and time it was given, and the course it is filed under. Each is a change you could make yourself, and it reaches your phone and the website like any other. It cannot change or delete anything else of yours.
• That app belongs to another company — OpenAI, Anthropic or whoever makes it — and what it does with what it reads is under that company's terms and privacy policy, not ours. Connecting it is your choice and your instruction to us to send it what it asks for.
• We keep which apps you connected — their names, where they send you back to, when you connected them and when they last read anything — and the connection's keys, stored only as hashes. Disconnect an app on your account page and it reads nothing more.

Sharing a lecture, if you do

With an account you can share one lecture as a link, from the app or the website. Apart from an AI app you connect yourself, this is the one way anything of yours becomes readable by someone else, so exactly what it does:

• Anyone who has the link can open it in a browser, without an account. What they see is that lecture's title, its course's name, its date, its transcript and its summary, and its audio summary if you made one. Never the recording — it is not on our server to show — and nothing else from your account: not your email, not your other lectures, not your questions.
• The link is a long random address that cannot be guessed, and the page asks search engines not to index it. Anyone you send it to can send it on.
• It shows the lecture as it is when it is opened, so a transcript you correct is corrected there too. Stopping sharing, deleting the lecture or deleting your account turns the link off at once, and the link stays dead: sharing again makes a new one.
• A lecture shared from 2 October 2026 on, where the Share panel says so, can also be added to the library of anyone with a Cathedra account who has the link. They then hold a copy of everything filed under that lecture — its transcript, its summaries, quizzes and mind maps, its flashcards, its translations, what it announced, its photographs and its audio summary. Never the recording, and never your questions, your quiz attempts or how well you know a card. The copy is theirs, in their account: stopping sharing, correcting or deleting the lecture afterwards does not change or remove it. A link made before that date is read and nothing more.
• A course can be shared as one link too — all of its lectures, or the ones you tick. Anyone who has that link sees the course's name and each of those lectures' title, date and length, and nothing more; with a Cathedra account they can add the lectures they choose to their own library, each as the copy described above — never the recordings. A link for the whole course includes lectures you record after making it, until you stop sharing it.
• To make it work we keep the link's address, which account and lecture or course it belongs to, which lectures it carries, and when it was made — nothing about who opens it.

Usage data

To see which parts of Cathedra are used and which are not, the app and the website count how they are used. Settings turns this off, on the phone or on the website, and turning it off deletes what was collected. While it is on:

• What is recorded is which screens are opened and which features are used, and when — a recording started, a summary made, a search run — together with the app or website version, the kind of device and its system version, the language it is set to, the country the request came from, a feature that failed and which one, and a subscription started or ended.
• It is recorded against a random identifier made on your device for this purpose alone. It is not your account, your email address or anything Apple gives apps, and we keep nothing that links it to them; a new one is made if you turn usage data off and on again.
• Two versions of a screen. Sometimes a screen — the subscription screen, say — is shown in one of two versions, to learn which works better. Your device picks the version itself, from that random identifier, and what is recorded is which version it showed, as one more entry of the kind above. Nothing about it is tied to your account, and with usage data off you are shown the ordinary version and are part of no comparison. The prices are the same in both.
• Never recorded: anything a lecture, a transcript, a note, a question or an answer says; the audio of anything; your email address; your IP address; any advertising identifier.
• It goes to our own server on Cloudflare, in Western Europe, and to no other company. It is not sold, not shared, not used for advertising and not combined with data from anywhere else.
• It is kept for twelve months and then deleted. Turning usage data off deletes what was recorded under that identifier, and the identifier with it, at once.
• You can object to it without turning it off by writing to us, and we will delete it and stop.

Deleting things

Deleting the app removes all of it, and the app lets you delete any recording, any transcript, any course or any term individually at any time.

The microphone

Cathedra asks for microphone access so it can record lectures. Recording only ever starts when you start it — from the app, from a reminder's Start Recording button, or from the Lock Screen card. iOS does not permit an app to open the microphone on its own, and Cathedra has no way to do so. iOS shows its own recording indicator whenever the microphone is live.

Speech-to-text

Transcription runs on your device, using the speech recognition built into iOS. Your audio is not sent to us, to Apple, or to anyone else in order to be transcribed.

One thing does involve the network, and it is worth being precise about: the first time you transcribe a lecture in a language you have not used before, iOS downloads that language's speech model from Apple. That download is a request to Apple for a model file. It carries no audio, no transcript and nothing about your lectures. After it completes, transcription in that language works with the network switched off entirely.

Notifications

Class reminders and exam reminders are scheduled locally by your iPhone from the timetable you entered. They are not push notifications; nothing about your schedule is sent anywhere. If you decline notification permission, the rest of the app works normally.

What we do not do

• No password. An account is an address and a code sent to it, so there is no password to store, to leak, or to reset.
• No name, no phone number, no date of birth. Nothing is asked for that the app does not use.
• No usage data beyond what is described above — counted by us, never by another company's analytics, and switched off in Settings whenever you like.
• No tracking: nothing about you is linked with data from other companies, and nothing is used for advertising.
• No advertising and no advertising identifiers.
• No third-party SDKs of any kind are included in the app. Every service above is reached over an ordinary web request to our own address; no outside company's code runs inside Cathedra.
• Nothing is ever sold. Nothing is shared with anyone except the companies named above, each doing one job we asked them to do, and an AI app you connect yourself.
• We never receive the audio of a lecture you record on your iPhone. Not on our server, not on the way to anywhere. The only sound that leaves the app is the few seconds of your own voice when you choose to speak a question instead of typing it; the only lecture audio on our server is a file you upload on the website yourself, described above.
• We do not read your library for our own purposes. What is on our server is there so the website can show it to you and answer your questions from it, and an AI app you connect can read it, and for nothing else — not to train a model, not to build a profile, not to look at.

Recording other people

This part is about your responsibility rather than ours, and it matters more than anything else on this page.

A lecture recording contains someone else's voice — usually your lecturer's, sometimes other students'. Whether you may record it is decided by the law where you are and by your university's rules, not by this app. Many institutions permit recording for personal study and prohibit sharing; some require the lecturer's permission first; some prohibit it outright. Rules differ between countries and between universities in the same country.

Please check before you record, ask when in doubt, and do not publish or distribute a recording of someone who did not agree to it. Cathedra keeps your recordings private on your device by default; what you choose to export or share afterwards is up to you.

Children

Cathedra is intended for university and college students and is not directed at children. If you believe a child has made an account, write to us and we will delete it.

Why we are allowed to hold it, and for how long

Under the GDPR every piece of personal data has to have a reason. Ours:

• Your email address, your sessions, and the copy of your library on our server — to give you the service you asked for when you made the account. Kept while the account exists. Delete the account and they go with it, immediately and completely.
• A sign-in code — the same reason. Stored only as a hash and gone after ten minutes, used or not.
• Sending lecture text, a photograph or a few seconds of your voice to be answered — to give you the answer you asked for, which is the service you subscribed to, and which cannot be given without sending what the question is about. Cathedra AI is off with one switch in Settings, and the rest of the app keeps working. We store nothing of what is sent beyond what your account already keeps; what each service keeps is under Who else sees any of it.
• Sending the phrases that sound misheard, when a lecture has been written down — to give you a transcript of what was actually said, which is what the app is for, and which a speech recogniser on its own does not manage for the technical words of a subject. It is a list of phrases rather than the lecture, we do not store it, and the same switch stops it.
• A shared lecture's or course's link — to give you the service you asked for when you shared it. Kept until you stop sharing, delete the lecture or delete the account.
• A connected app — to do what you told us to when you connected it. Kept until you disconnect it, delete the account or leave it unused for ninety days; the key it reads with lasts an hour at a time.
• Sending a lecture to be made into an audio summary, and keeping that summary with your account — the same reason: it is what you asked for, and it is made by the two services named above. The summary is kept while the account exists, or until you make another of the same lecture, which replaces it.
• Usage data — our legitimate interest in knowing which parts of Cathedra are used and which are not, and which of two versions of a screen works better, weighed against data that carries no name, no address and nothing a lecture says. Kept for twelve months, switched off in Settings — which deletes it — and objected to by writing to us.
• What you agreed to, and when — so that we can show, if we are ever asked, what you accepted and chose, which the law requires of anyone relying on a consent. Kept while the account exists, and deleted with it.
• Your subscription's identifier and this month's question count — to check that a subscription is live and to keep a fair-use ceiling, which is our legitimate interest in not being billed for somebody else's use of a leaked identifier. The identifier is kept for a few hours at a time; the count for the month it belongs to.

Who else sees any of it

Five companies, each doing one job. None of them is given anything for its own purposes, except as said of DeepSeek below:

• Cloudflare — runs our server and holds the account, the copy of your library and, if you turned it on, usage data, in Western Europe.
• DeepSeek — answers a question, from the text sent with it. In the People's Republic of China: it stores what it receives there, keeps it in a cache for hours to days so that a follow-up question costs less, and under its own terms may use it, to a limited extent, to improve its services. It does not offer the European Union's standard contractual clauses.
• Groq — turns a spoken question into words, and transcribes a lecture you upload on the website. In the United States; its terms do not let it train on what it receives.
• Google — speaks an audio summary's conversation aloud, through its Gemini API on terms that do not let it use what is sent to train its models; Google keeps it for up to 55 days to detect abuse. In the United States.
• Apple, and Paddle if you subscribed on the website — sell the subscription and tell us only whether it is live. Neither gives us your payment details.

An AI app you connect to your library is not on this list, because it is not ours: it reads what you let it, under its own company's policy — see Your own AI app, if you connect one.

One page on this site, the payment page, loads Paddle's checkout script from Paddle's own servers and shows Paddle's payment form inside it. What you type there goes to Paddle, not to us, and what Paddle keeps is described in Paddle's privacy policy. No other page here runs a script.

The three that sit outside the European Union and Türkiye receive what a feature you use needs, at the moment you use it — text, a photograph or a few seconds of your voice — and DeepSeek also receives the phrases that sound misheard when a lecture has been written down. Nothing about you is sent with it beyond what that needs.

Your rights

Data protection law — including the GDPR in the European Union — gives you the right to see what is held about you, to have it corrected, to have it deleted, to take it elsewhere, to object to it being held, and to withdraw a consent you gave. You can use any of them by writing to support@cathedra.study, and we answer within a month.

Most of it you can do yourself, without asking anyone. Deleting your account — in the app under Settings › Account, or on the website — erases the address, every session and the whole copy of your library, and it happens as you watch. Signing out stops the copying. Turning Cathedra AI off in Settings — in the app or on the website — stops anything being sent, and turning usage data off stops it being counted and deletes what was.

Your recordings are on your own device, under your own control: the app can play, export in several formats, and permanently delete any of them, and deleting the app removes everything at once. We could not give you those or delete them for you if we wanted to — we have never had them.

If you think we have handled your data wrongly, tell us first and we will try to put it right. You also have the right to complain to a data protection authority — in the Netherlands the Autoriteit Persoonsgegevens, and otherwise the one for the country you live in.

Changes

If this policy changes, the updated version will be posted at this address with a new date at the top. You accept this policy and the terms by entering your email address and going on when you sign in, and by using the app. When a change affects what you agreed to, Cathedra asks you to accept it once — in the app or on the website, whichever you open first — and keeps your answer with your account, so the other does not ask again. If you decline, you are signed out there; recording, transcription and search in the app keep working without an account.

Contact

Questions about this policy, or about the app: support@cathedra.study