Effective 2026-07-07 · updated 2026-10-18 · Terms of Service · support@vitalinguist.com

Privacy Policy

vitalinguist is a language-learning app. It works by having you converse with an AI in the language you're learning and by letting you practice against recorded material (films, audiobooks, coach-uploaded courses). To do those two things well, the app needs to know some things about you.

This document lists exactly what the app and its server (va.vitalinguist.com) collect, why we collect it, how long we keep it, and what you can do about it.

The 30-second version

• We do not sell your data. Ever. To anyone.
• We do not run third-party analytics or ad SDKs. There are no tracking cookies, no Facebook / TikTok / Google pixels — with one exception: the Meta pixel runs on the two free pages /sandwich and /unchew, and nowhere else. See “Advertising on the free pages” below.
• The only outside companies that receive what you say or write are the AI services that understand and answer it: Anthropic (Claude models), Groq and OpenRouter. They are in the United States. We ask you before the first time anything is sent, you can turn it off in Settings, and they see only the words, never your identity. See “Third parties” and “Where your data goes” below.
• You can delete your account at any time and everything private about you is purged — recordings and uploaded files included, not just the database rows. Reports you made public and recordings of group and one-to-one sessions stay, with your name removed; see “What stays when you delete your account”. The other exception is your attempts at phrases we gave you, kept without the audio and with the link to you severed, so the scoring keeps improving; see “What we keep to improve the app”.

What we collect

Account identity

• Email address — for magic-link login. Stored so we can send you a login code when you ask for one.
• Server-issued student id — a number we assign to your account. Held on your device in a system keychain; sent with every request as a bearer token.
• Telegram user id — optional, only if you signed in via a Telegram bot.

We use these to know it's you. Nothing more.

Voice recordings you send us

When you record a practice attempt or a Trainer voice turn, the audio file is sent to our server and transcribed by our speech-to-text service. The transcript is stored so you can see it back inside the app, and the recording itself is kept so you can hear your own past attempts — on a lesson phrase, in a story, or when you practise a grammar structure or check a phrase in Activation. You can delete any single recording from the attempt it belongs to, at any time; the transcript and the feedback stay. If you delete your account, every practice recording of yours is deleted with it — the audio files themselves, not just the links to them. Optional session-video recordings you explicitly start are stored so you can review them.

Your Trainer conversations

The text of every message you send to the Trainer and every reply it sends back is stored so the Trainer can remember what you were just talking about. This data lives in our database and is not shared with anyone.

The current conversation's most recent turns are sent to the LLM providers (see “Third parties” below) so they can generate the next reply. They receive the conversation contents; they do not receive your email address, your student id, or any information that identifies you personally.

Your vocabulary and progress

• Words you've marked “I know this” (known_words).
• Words the AI has introduced to you (trainer_vocab_log — how many times you've been exposed, whether you've used the word back).
• Recurring language patterns the AI has noticed you're working on (trainer_error_patterns — coarse categories like “articles” or “verb tense”, never public examples).
• A short summary the AI writes about you to keep the conversation personal across sessions (trainer_student_summary).

All of this stays on our server. It's what makes the Trainer a partner and not a generic chatbot.

Optional intake questionnaire

If you filled out the intake questionnaire, the answers are stored so the Trainer can reference your background naturally in conversation (“things you're proud of”, “films you keep returning to”). You can skip the questionnaire entirely.

Reports you post

If you post progress reports in the app, the text is stored and visible to your coach and to other students in the feed as you configured. You can delete any report you posted at any time. If you delete your account, a report you made public stays, with its recording and replies, and the name on it reads “deleted”.

Usage minutes (heartbeat)

The app pings our server once a minute while you're actively using it so we can show you a “you practiced X minutes today” counter. We store the per-minute bucket. We do not store what you were doing during that minute.

What we do NOT collect

• Location. The app never asks the OS for your location.
• Contacts. Ever.
• Photos other than the video you explicitly choose to upload as part of a report.
• Advertising identifiers (IDFA, GAID).
• Third-party cookies. We ship none, except the Meta pixel’s own on the two free pages. If you visit vitalingui.st in a browser we set exactly one session cookie for login state.

Permissions on your device

• Microphone — to record your voice for practice attempts and Trainer voice turns. Only when you tap the mic button.
• Camera — to record your face during a session, if you choose to. Only when you tap the record button and confirm.
• Photo library — to save an optional session video to your camera roll and to pick a video to attach to a report.
• Notifications — to tell you when your coach has answered a session of yours.

Every permission is off by default and requested only when you first use the feature that needs it. Denying a permission means the corresponding feature won't work; nothing else breaks.

Third parties

You are asked, in the app, before any of the three AI services below receives anything, and you can turn it off again at any time in Settings. If you say no, nothing is sent to them: reading, listening, writing and the exercises that mark themselves keep working, and only the checking of what you said or wrote stops.

• Speech-to-text. Turning your recording into text is normally done on our own machines. Only if they are unavailable is the recording sent to an outside speech-recognition service — Groq, or OpenRouter as a last resort. It receives the audio only, not your name or email.
• Anthropic (Claude models, reached through OpenRouter) — generates Trainer replies and checks what you said or wrote. Receives the current conversation's most recent turns plus your intake questionnaire highlights. Does not receive your email, student id, or any identifying detail.
• Groq / OpenRouter — the same role, also used as fallbacks and for the checking. Same data envelope.
• Meta Platforms Ireland — only on the free pages /sandwich and /unchew, through the Meta pixel. See “Advertising on the free pages” below.
• Firebase Cloud Messaging — Google's push notification service, used to deliver booking reminders. Receives a device push token; does not receive your email or Trainer conversations.

None of these services is permitted to keep what we send or to train its models on it. Every request we make to OpenRouter carries an explicit instruction to route only to providers that do not collect the data; Groq's own terms bar training on what its API is sent and retain nothing by default. We do not send anything a student wrote or said on a personal subscription account, which would carry different terms — that is enforced in the code, not by convention.

We do not use analytics platforms (no Firebase Analytics, no Google Analytics, no Mixpanel, no Amplitude, no Sentry).

Why we may use your data

The law (GDPR, Article 6) lets us use personal data only on a stated ground. Ours are:

• To give you the app you signed up for (contract, Art. 6(1)(b)): your account, the Trainer, your recordings and progress, sign-in, reminders you asked for. Without this data the app cannot work.
• To send what you say or write to the AI services (your consent, Art. 6(1)(a)): you are asked first and can withdraw at any time in Settings; withdrawing does not undo what was already checked.
• To keep the service safe and working (our legitimate interest, Art. 6(1)(f)): server logs, preventing abuse, and the anonymised phrase attempts described below.
• To keep the records the law requires, such as invoices (legal obligation, Art. 6(1)(c)).

Giving us your email and using the app is your choice; if you do not, we cannot open an account. We take no decisions about you by automated means that have legal or similar effects: the AI only checks your practice and writes replies.

Where your data goes

Our servers are in the European Union. The three AI services above are companies in the United States, so what you say or write, once you have agreed, leaves the EU. For this we rely on the European Commission’s standard contractual clauses (GDPR Article 46), which are part of those companies’ data-processing terms. Meta Platforms Ireland and Google (push messages) may also process data outside the EU under their own safeguards. To ask how a safeguard applies to you, write to support@vitalinguist.com.

Advertising on the free pages

The two free pages /sandwich and /unchew carry the Meta pixel, so we can see which of our ads help people and show them to the people they help.

Who receives the data: Vitaly Zubkov (Vitalinguist, Spain) and Meta Platforms Ireland. What is passed on: your IP address, your device and browser type, which of these pages you opened, whether you finished the exercise and pressed the button to buy the week, and, if you came from one of our ads and then pay, the ad click number and the amount you paid. Your email is never passed on. Why: to measure and improve our ads; Meta may also use this data for its own advertising. On our side, the mark of the ad you came from is kept in your browser for up to 30 days. A browser’s tracking protection or an ad blocker stops the pixel; the exercise works the same either way.

How long we keep your data

• Account data, Trainer conversations, vocabulary, progress: for as long as your account exists.
• Delete-account request (how to ask, including if you can't sign in): we delete it. Your account details, your Trainer conversations, your recordings, your uploads, your written practice, your goals, your vocabulary and your progress are removed, not hidden. Two kinds of thing stay without your name — see “What stays when you delete your account” — and one narrow anonymous set, see “What we keep to improve the app”.
• Voice recordings: kept so you can hear your past attempts, until you delete them. You can delete any single recording from inside the app at any time, and all of them are deleted — files included — when you delete your account (except the recording on a report you made public).
• Recordings of group and one-to-one sessions: kept, including after you delete your account, with your name removed.
• Videos you explicitly recorded of yourself in the app: kept until you delete them (or delete your account).
• Server access logs: rotated after 30 days.

What we keep to improve the app, and why

The app scores your speaking two ways. Some of it is deterministic — ordinary code comparing what you said against the phrase you were given, word by word. The rest uses AI models. Both get better by being measured against real practice rather than invented examples, because the mistakes learners actually make are not the ones we would think to write down.

So when you delete your account we keep one narrow category: your attempts at phrases we gave you — the words we asked you to say, what our speech-to-text heard back, and the score. Before it is kept it is anonymised, and we mean the strict version of that word:

• the link to you is severed, not merely relabelled — the rows cannot be gathered back together into one person, so there is no “anonymous user #47” whose practice history can be read as a whole;
• the audio is deleted. A voice recording identifies the person speaking, and no amount of relabelling changes that, so recordings are never part of this anonymous set;
• the text is a phrase we wrote and asked you to repeat, not something you composed about yourself.

Everything else goes: anything you wrote privately, your practice recordings, anything you uploaded, your conversations with the Trainer, your goals and your progress — except what is listed under “What stays when you delete your account”.

What stays when you delete your account

• Reports you made public for other students to see, with their recording and the replies on them. They stay where they are; the name on them reads “deleted”.
• Recordings of group and one-to-one sessions. Members agree to these being recorded and stored when they join. We keep them, with your name removed, to study how people learn. After you delete your account we no longer use them in our marketing.

A voice or a face can still be recognised, so these are not anonymous — they are kept without your name, and we say so plainly. Nothing private of yours is in this list.

Your rights

• Access: download everything we hold about you, from Settings, whenever you like — you don't have to ask anyone. Sign-in tokens and codes are shown as present but their values are withheld, so the file is not a way into your account. If you can't sign in, email support@vitalinguist.com.
• Correction: edit your intake questionnaire from Profile at any time, or ask support for anything else.
• Deletion: Settings → Delete account from within the app. Confirmation is required; the operation is not reversible.
• Objection, restriction, and withdrawing consent: you can object to processing based on our legitimate interest, ask us to restrict it, and withdraw your consent to the AI services at any time in Settings. Write to support@vitalinguist.com; we answer within one month.
• Portability: that same download is machine-readable JSON — your account, every table of your own work, and an index of your recordings by address. Recordings are listed rather than embedded so the file stays something you can open.

If you're in the EU or UK, GDPR/UK-GDPR rights apply and support@ is the right contact. If we can't resolve your concern you can complain to your national data protection authority; in Spain that is the Agencia Española de Protección de Datos (aepd.es).

Children

The app is not directed at children under 14, the age from which a person in Spain can agree to the use of their own data. We do not knowingly collect personal data from anyone under 14. If a parent or guardian believes we have data about a child, email support@vitalinguist.com and we'll delete it.

Changes to this policy

We'll email you (using the address on file) and post an in-app notice at least 30 days before any change that expands what we collect or how we use it. Non-material tweaks (wording, links, contact info) may land without notice.

Contact

support@vitalinguist.com
Vitaly Zubkov (individual entrepreneur), Madrid, Spain — the controller of your data. We have no data protection officer.