Privacy Policy
Effective date: 2026-10-05 Last updated: 2026-10-05
1. Who is responsible for your data
The controller of your personal data is Pavel Slapovskii, trading as Studify, Vojvode Maša Đurovića, 81000 Podgorica, Montenegro — the operator of Studify (https://studify.today, app at https://app.studify.today). Contact: support@studify.today.
Payments are separate. Lemon Squeezy, LLC is the merchant of record for purchases. It collects your billing details and payment method directly and acts as an independent controller for that transaction data under its own privacy policy. We never receive or store your card number. We receive only what we need to unlock your plan: your purchase email, the plan you bought, subscription status and dates, and an internal customer and subscription identifier.
2. What we collect
| Category | Examples | Where it comes from |
|---|---|---|
| Account | email address, Google account identifier if you sign in with Google, sign-in timestamps | you, when you sign in |
| Lesson content | transcripts of what you said during a lesson, the mistakes detected, vocabulary you practised, your progress level, the tutor's replies, the lesson report | generated during lessons |
| Voice recordings | short audio fragments of pronunciation exercises (see section 4) | your microphone, during pronunciation exercises |
| Usage | lesson start and end times, duration, which features ran (speech recognition, synthesis, avatar, pronunciation scoring) and for how long | generated automatically |
| Technical | session cookie, IP address and browser data in server logs, error reports | your browser |
| Billing | plan, subscription status, renewal dates, purchase email, Lemon Squeezy customer identifier | Lemon Squeezy |
| Product analytics (only if you agree) | which screens you opened, whether a lesson started and finished, why one could not start, that a purchase was made — with the plan code, never the amount | generated automatically while you use the App |
| Referral source | the campaign tags of the link you arrived through (the utm_… parameters visible in the address) and, if the platform that hosted the link added one, its advertising click identifier — recorded the first time you sign in and on later sign-ins that arrive through a tagged link |
the link you clicked |
| Advertising events (only if you agree) | the fact that you started a free lesson, and the fact that you paid together with the amount — each sent with a scrambled (hashed) form of your email address and the advertising click identifier of the link you arrived by | generated when those things happen |
Product analytics never contains your voice, what you said, what the tutor replied, or your email address — only which of the events above happened and when.
Your referral source is a few short labels about a link, not about you: which account, post or
campaign it belonged to. We keep it to know which of our own posts and channels bring people who
actually learn with us. Which exact post you came from stays with us: what we send to our analytics
processor is the source and traffic type only (for example threads and social).
The advertising click identifier stays with us too — unless you give the advertising consent in section 3. With it, and only with it, we tell the advertising platform named in section 6 that a free lesson was started or a payment was made, and we send the click identifier and the hashed form of your email so that the platform can match the event to the ad you saw. We send nothing else: not your voice, not what you said, not your lesson results, not your email address itself, and not which post you arrived from.
We do not collect special-category data intentionally and we do not sell personal data. We build no advertising profile of you ourselves; the advertising platform does its own profiling under its own policy, and only if you agreed.
3. Why we use it, and on what legal basis
| Purpose | Legal basis (GDPR Art. 6) |
|---|---|
| Running a lesson: recognising your speech, generating the tutor's replies, showing corrections | performance of a contract (Art. 6(1)(b)) |
| Keeping your progress, mistake history and lesson reports so lessons build on each other | performance of a contract |
| Storing and scoring pronunciation audio fragments | your explicit consent (Art. 6(1)(a)), which you can withdraw at any time |
| Understanding how the App is used, so we can find and fix what does not work | your consent (Art. 6(1)(a)), which you can withdraw at any time |
| Measuring advertising: telling the advertising platform that a sign-up or a payment happened, so we can tell which ads work | your consent (Art. 6(1)(a)), which you can withdraw at any time |
| Billing, plan entitlement and fraud prevention | contract and legal obligation |
| Security, debugging, reliability and measuring latency | legitimate interests (Art. 6(1)(f)) |
| Knowing which channel brought you, so we can keep what works and stop what does not | legitimate interests (Art. 6(1)(f)) — the data is a handful of labels about a link, we build no advertising profile from it, and you can object at any time |
| Account emails (sign-in links, receipts, plan changes) | contract |
You give these consents on the screen you see before your first lesson, and you withdraw any of them in Your data in your account — one switch each, no email to us required. They are optional and independent of one another: refusing any of them changes nothing about the lessons you get.
Withdrawing the audio consent stops pronunciation scoring and also deletes the fragments we already hold, along with the scores calculated from them. Withdrawing the analytics consent stops the counting and deletes what was already collected about you.
Withdrawing the advertising consent stops any further events from being sent, from the next one onwards. It cannot recall what was already sent: that data is held by the advertising platform under its own rules, and we have no way to reach into it event by event. To have it removed along with everything else, delete your account (section 8). Withdrawal does not make processing that already happened unlawful.
4. Voice: what we actually keep
This is the most sensitive part of the App, so we are specific about it.
- Your speech is streamed to a speech-recognition provider and turned into text in real time in order to run the conversation. We do not keep a recording of the whole lesson.
- Only short audio fragments of pronunciation exercises — typically a few minutes across a lesson — are stored, because they have to be scored and played back to you as feedback.
- The transcript-derived material we keep (mistakes, vocabulary, the report) is text, not audio.
- Nothing is recorded at all unless you gave the optional consent. Without it the exercises still run; your voice is simply never written down anywhere.
- Audio fragments are stored in the EU and deleted after 90 days by a job that runs every day — not on request, and not when someone remembers. They go sooner if you withdraw consent or delete your account.
5. Where your data is stored
All durable personal data is stored in the European Union:
| What | Provider | Region |
|---|---|---|
| Accounts, profile, lessons, mistakes, progress | Supabase (PostgreSQL) | Ireland (eu-west-1) |
| Audio fragments and lesson reports | Cloudflare R2 | EU |
| Pronunciation scoring | Microsoft Azure Speech | North Europe (Ireland) |
| Live session state during a lesson (temporary, expires automatically) | self-hosted Redis on a DigitalOcean server | London, United Kingdom |
| Product analytics events (only if you agree) | PostHog Cloud | Frankfurt, Germany |
The live-session cache in London holds only the state of a lesson while it is running and expires by itself; nothing durable is kept there. The United Kingdom is covered by a European Commission adequacy decision, so this transfer needs no additional safeguards.
Some processors listed below are based outside the EU (mainly the United States). Where that is the case, transfers are made under the European Commission's Standard Contractual Clauses and/or the EU–US Data Privacy Framework, as provided in each provider's data processing agreement.
6. Who processes data for us
| Processor | What it does | Data involved |
|---|---|---|
| Lemon Squeezy (merchant of record) | payments, tax, receipts, subscription management | billing data, purchase email |
| Supabase | database and authentication | account, lessons, progress |
| Vercel | hosting the web application | technical data, request logs |
| DigitalOcean | server running the lesson agent | technical data, live session state |
| LiveKit | real-time audio/video transport of the lesson | live audio/video stream |
| Deepgram | speech recognition and speech synthesis | live audio, generated speech |
| OpenAI | the conversational model driving the tutor | lesson text sent to the model |
| Anthropic | offline model that writes lesson reports and materials | lesson text sent to the model |
| Microsoft Azure Speech | pronunciation assessment | audio fragments of exercises |
| Simli | photorealistic avatar video (premium mode only) | generated speech audio, avatar video |
| Cloudflare R2 | storage of audio fragments and reports | audio fragments, reports |
| Resend / Amazon SES | transactional email (sign-in links, notifications) | email address |
| Langfuse | technical tracing of the AI pipeline | in production, metadata only — timings, token counts and model names; the text of speech is stripped before it leaves our servers |
| PostHog (EU region) | product analytics: which screens are opened, whether lessons start and finish, where people get stuck | only if you agree — the events listed in section 2, tied to your account identifier; never your voice, your answers or your email |
These providers act on our instructions under data processing agreements. We do not permit them to use your data to train their models, and we do not sell or rent your data to anyone.
Meta is different, and we say so plainly. If you give the advertising consent, the events in section 2 go to Meta Platforms Ireland Limited. For those events Meta is not our processor: under its Business Tools Terms we and Meta are joint controllers for the collection and transfer, and Meta is then the sole controller of what it does with the data afterwards — measurement, audiences and improving its own advertising systems — under its own data policy. What reaches it is the hashed form of your email, the advertising click identifier, and the fact and amount of the event; nothing from your lessons. You can withdraw the consent at any time in Your data, and you can exercise your rights against Meta directly as well as against us.
7. How long we keep it
| Data | Retention |
|---|---|
| Account, profile, progress, mistakes, vocabulary | while your account exists |
| Lesson reports | while your account exists |
| Pronunciation audio fragments | 90 days, then deleted automatically |
| A copy of your data you asked us to prepare | 7 days, then deleted automatically |
| Live session state | automatically expires shortly after the lesson |
| Product analytics events | while your account exists and your consent stands; deleted when you withdraw it or delete your account |
| Referral source | while your account exists; deleted with it |
| Advertising events sent to Meta | held by Meta under its own retention rules, outside our control — we keep no copy of the event beyond what is already listed above |
| Billing records held by Lemon Squeezy | as required by tax and accounting law, independently of us |
The two automatic deletions above are done by a job that runs daily; they do not depend on anyone noticing that a deadline has passed.
Deleting your account deletes the data we hold about you, including your sign-in credentials — you do not stay in our systems as an empty account. Financial records that Lemon Squeezy must keep by law are outside our control and are retained by them.
8. Your rights
If you are in the EU/EEA or the UK you have the right to access your data, correct it, delete it, restrict or object to processing, withdraw consent, and receive a portable copy. You also have the right to complain to your national data protection authority.
Four of these you exercise yourself, without asking us, in Your data in your account:
- Withdraw consent for audio. One switch. It stops pronunciation scoring and deletes the fragments we already hold, along with the scores derived from them.
- Withdraw consent for analytics. One switch. It stops the counting and deletes what was already collected about you at our analytics provider. Your lessons are unaffected either way.
- Get a portable copy. We put together a single JSON file with everything we hold about you — profile, lessons, mistakes, vocabulary, pronunciation scores, purchases and your consent history — and give you a download link. It does not contain audio of your voice; if you want the stored fragments themselves, ask us by email while your account still exists. The file itself is personal data, so it is available for 7 days and then deleted.
- Delete your account and data. You confirm by typing your own email address; we then sign you out and delete your lessons, reports, recordings, vocabulary, purchase records and your account itself. This usually takes minutes. It cannot be undone, and lessons you paid for but did not use are lost.
For everything else — access, correction, restriction, objection, or a complaint about how any of this worked — email support@studify.today from the address on your account. We respond within 30 days.
Two warnings worth reading before you delete anything. Deleting your Studify data does not cancel a subscription — cancel it in the customer portal first, or ask us and we will do it, otherwise you keep being charged for lessons you can no longer take. And while you can sign up again later with the same email address, your previous lessons, reports and progress cannot be restored.
9. Automated decision-making
The tutor is an AI system: it decides in real time what to say, what to correct and how to score your pronunciation. These are learning judgements — they do not produce legal effects and do not significantly affect you in the sense of GDPR Art. 22. They can be wrong; you can always disagree, ask the tutor to move on, or write to us.
10. Cookies and storage on your device
We use two cookies, and both are strictly necessary for the App to work:
- a session cookie, which keeps you signed in (plus, where applicable, a security cookie for the sign-in flow itself);
- a small consent cookie, which records that you have accepted the current version of these documents, so we do not stop you at the consent screen on every page you open.
Neither is used for advertising, and neither is shared with anyone.
Product analytics stores nothing on your device at all — no cookie, no local storage, no device fingerprint. Events are tied to the account you are already signed in to, and if you never agree to analytics, nothing is sent in the first place.
The public website counts visits, and it counts them anonymously. Each time a page of the public
website is opened, it adds one to two kinds of counter kept per hour: one by the broad kind of place
you came from, and — only if the link carried campaign tags, the utm_… parameters visible in the
address — one by that set of tags.
The "kind of place" is one word from a fixed list we chose in advance (a search engine, an AI assistant, one of a few social platforms, "direct", or "other"). The address you came from is never sent to us: your browser reduces it to that single word before anything leaves it, so the page you were on and anything you typed into a search box stay out of our systems entirely.
Neither counter contains an identifier of any kind: no cookie, no local storage, no device fingerprint, no IP address, nothing about your device or browser. They cannot be linked back to you, and they are the same counters whether or not you later create an account. Because nothing is placed on your device and nothing in them identifies a person, this counting does not depend on any of your consents — and switching a consent off does not switch it on either. (As with any request to any website, your IP address appears briefly in server logs — see section 2.)
Lemon Squeezy sets its own cookies on its checkout pages under its own policy.
11. Children
The App is for people aged 18 and over. We do not knowingly collect data from children. If you believe a child has created an account, write to support@studify.today and we will delete it.
12. Security
Access to production data is limited to the operator. Data in transit is encrypted (HTTPS/WebRTC); data at rest is encrypted by our storage providers. Secrets are held in managed secret stores, not in code. No system is perfectly secure; if a breach affects your data we will notify you and the relevant authority as required by law.
13. Changes
We may update this policy. Material changes will be announced by email before they take effect. The current version is always at https://studify.today/privacy.
14. Contact
Pavel Slapovskii — support@studify.today — Vojvode Maša Đurovića, 81000 Podgorica, Montenegro.