Privacy Policy

Learn with Alphonso is a language learning app for English, French and Spanish, operated by Shayan Salimi ("we", "us"). This policy explains what we collect, why, who else handles it, and the choices you have. It covers the iOS app, the Android app where it is available, and the website at learn.alphonsoecosystem.app.

We are the data controller for everything described here. Write to support@alphonsoecosystem.app about anything in this policy or for help using the app. A person reads every message.

• Account details: your email address, how you sign in (email, Apple or Google), and your display name. If you sign in with Apple or Google, we receive the email address they share with us (with Apple this may be a private relay address) and, if they share it, your name. We use your first name only to suggest a display name. Apple shares your given name only the first time you sign in with Apple, so we keep it in your account record for that suggestion. We also store your profile picture (one of 16 preset animal pictures you pick in the app, or your initial on a generated colour; you cannot upload a photo), your chosen theme, and, only if you turn on Show my flag on leaderboards, your country as a two-letter code on your public profile.
• About you, when you start: the first name you want Alphonso and Hector to call you (optional), the language you speak, the country you are joining from, why you are learning, how much you think you know, how many minutes a day you want to study, your age range (for example 25–34; we store an age range, never your date of birth), the topics you enjoy (optional), and how you heard about us (optional). Until you create an account these answers stay in your browser; when you do, we save them to your account record. We use them to set up your course, your daily goal and your plan, and to choose topics for practice and conversations. Other learners never see them, except your country flag if you turn it on.
• Learning data: lessons completed, quiz and placement answers, XP, streaks, hearts, achievements, league and season standing, team membership and your team’s weekly mission progress (how many lessons you add to it, and the XP it pays out), duel results, your spaced-repetition review items, a learning goal if you set one (the level and date you choose), and any words you choose to save (the word, the sentence you tapped it in, and the meaning and answer choices we generate for it).
• Conversation practice: the text of messages you exchange with Hector and in practice scenarios and campaigns, your written answers to translation questions, and audio you record while speaking. These are sent to our AI providers to produce a reply, a transcript or a grade, and are not kept on our servers afterwards. What we do keep from a conversation is described in the Hector section below.
• Practice items from your mistakes: after a conversation, our AI looks for mistakes you make often. If it finds one, we save a short practice item to your review queue (a label for the weak spot, a question, answer choices and an explanation) and a record of the type of mistake.
• Social data: your friends and friend requests, study buddy pairings and the preset messages you exchange, whether you are waiting for a matched study buddy and that you confirmed you are 13 or older to use matching, team membership and any team name you choose, duel invitations, learners you block, and reports you make or that are made about you.
• AI consent: whether you have allowed AI features on your account, and when.
• Audio library use: which episodes you play, how far through you are, and which you download for offline listening. Episodes marked AI-narrated are read by a synthetic voice (Deepgram text-to-speech) from scripts we write; making them uses no personal data.
• Purchases: whether you hold an active Alphonso Pro subscription, where you bought it, and its status and renewal dates. We never receive or store your card details. Apple, Google Play on Android, or Stripe for a subscription bought on our website, handles payment.
• Notifications: if you enable push notifications, a device token so we can send them. It is used for nothing else.
• Usage limits: counts of how many AI requests your account makes each day, so we can apply fair-use limits.
• Speaking time: a daily total of how many seconds you spoke in practice, so we can show your progress. We keep only the number, never the audio or what you said, until you delete your account.
• Learning time, skills and listening: a daily total of how many seconds you spent on lessons and practice, which of five skills (listening, speaking, vocabulary, grammar, reading) each answer counted towards, a marker for each saved word you have mastered, your time zone (used to show dates and to decide which calendar day your streak and daily totals count towards; if we do not have it, we use UTC), and how far through each audio episode or video you are (and when you finished a video), so we can show your progress and let you continue listening or watching. We keep these only as totals and positions, never audio or what you said, until you delete your account, and they are in your data export.
• Actions by our team: if our staff give you free Pro, block you from social features, suspend your account, add a private note, or delete your account, we keep a record of the action (who did it, what, when, and the reason). The record holds identifiers and short reasons, not your lessons, answers or chats; it is not part of your data export and, so that actions stay accountable, it is kept after your account is deleted.
• Where you came from: if you chose Accept all in the cookie banner and arrived through a link that carries campaign tags (utm_source, utm_medium, utm_campaign), those tags and the time you arrived, saved once when you create your account. For this we do not store your IP address, device, browser or the page you came from.
• Technical data: request logs and error reports needed to keep the service running and secure, including IP address, device or browser type, and the time and address of each request.
• Crash and error reports: if the app or website crashes or hits an unexpected error, a report is sent to Sentry with the technical details of the fault (the error message, a stack trace, the app version, your device model and operating system or browser, and the page or screen). Reports are not linked to your account, carry no email address, name or sign-in token (we remove them before sending), and include no screenshots or screen recordings. The iPhone and Android apps send them from released builds only.
• Anonymous usage data (optional, off until you choose it): if you choose Accept all in the web cookie banner, or turn on Share anonymous usage data in Settings (in the iPhone and Android apps, in the first-run choice or in Settings), we record which features are used, such as that a lesson was started or finished (with its level, course and a score band), a practice call was made, an episode was played, a video was watched, Hector or the Pro offer was opened, or a first-run step was reached, with the app version and platform. It never includes your name, email, account identifier, what you type or say, or AI chat content, and it is not linked to your account: it uses a random identifier that is reset when you turn the setting off. Nothing is sent before you choose, and never from a device where you said you are under 13.

When you first sign in, the app asks what other learners should call you. It suggests your first name if Apple or Google shared it with us, or otherwise a generated name like Learner-4F2A. Apple shares your given name only once, when you first sign in with Apple, so we keep it in your account record to make that suggestion. You can type a different name, or tap Skip. Skip gives you a generated name like Learner-4F2A, replacing any name that came from your Google account, and you can change it later. Your display name is visible to other learners, so you do not have to use your real name, and you should never include contact details in it.

Every name is checked against a filter for offensive words. If you signed up before this step existed and your display name was taken from your email address or did not pass the filter, we replaced it with a generated name, and the app asks you once to choose your own.

AI features (Hector, practice scenarios and campaigns, AI grading of written translations, speech transcription, spoken replies, word meanings and practice items from your mistakes) only run after you allow them. The first time you open one, the app explains what is sent and to whom, and asks for your permission. Your choice is saved on your account with the time you made it, so it applies on every device you sign in on.

Explain in my language: when you tap “Explain in” your own language (or choose “Always show” in Settings), the app sends the item to the AI, which is the question, your answer and the right answer, or the word and its sentence, or the chat message, together with the language you chose. The explanation is written in that language and labelled as AI. We keep each explanation in a shared store so the same item in the same language is only written once; it is not tied to your account, and another learner who asks for the same item in the same language may see it. Because an explanation can quote your answer, it may contain a few of your words. Your language is read from your profile, never sent by the app.

You can withdraw your consent at any time, in the iOS app under Profile → Settings → AI features, or on the website on your Profile page. From that moment nothing more is sent to our AI providers. Without consent you can still do lessons, review and the placement test: written answers are checked against the expected answers instead of by AI, and speaking questions let you type your answer. Withdrawing consent does not delete what was already saved, such as practice items in your review queue. Deleting your account deletes them. If our safety filter blocks part of an AI reply, a short excerpt of it (up to 120 characters) is written to our server request logs so we can check the filter.

Some extra practice questions are generated from the lesson itself. Those requests contain only the lesson and the course, and nothing about you, so they do not need your consent.

• To provide the service: saving your progress and syncing it across devices (performance of our contract with you).
• To operate leaderboards, leagues, friends, teams, duels and study buddies, which show your display name, profile picture and progress to other signed-in learners, and on leaderboards your country flag if you turned it on (performance of our contract).
• To run AI features: generating replies, transcribing your speech, grading written answers, writing word meanings and creating practice items from your mistakes (your consent, which you can withdraw at any time).
• To manage your subscription and unlock Alphonso Pro (performance of our contract).
• To send sign-in codes, password-reset links and other account emails (performance of our contract).
• To send push notifications you have enabled (your consent, which you can withdraw in your device settings at any time).
• To review reports, enforce our Terms of Use and keep learners safe (our legitimate interests and our legal obligations).
• To prevent abuse and control cost, including per-account daily limits on AI features (our legitimate interests).
• To count which links and communities bring new learners, using the campaign tags above (your consent, given by choosing Accept all; if you choose Essential only nothing is saved). The tags are deleted with your account.
• To keep the app secure and diagnose faults (our legitimate interests).

We do not sell your data, we do not share it with data brokers, and we do not use it for advertising. There are no advertising SDKs in our apps. The only SDKs that report on how the app runs are Sentry’s crash reporter and, only if you opt in, PostHog’s anonymous usage analytics, both described above.

Leaderboards, leagues, friends, teams, duels and study buddies show your display name, profile picture (the animal you picked, or your initial) and progress (such as XP and lesson counts) to other signed-in learners. Leaderboards also show the country you chose, but only if you turn on Show my flag on leaderboards; it is off unless you do, and you can turn it off in Profile at any time. Other learners cannot read the rest of your profile, including the answers you gave when you started.

A team's weekly mission shows every member the team's total progress and your own count of lessons. The mission does not list other members' individual counts. On a two-member team, while the mission is in progress, the total and your count reveal the other member's count.

If you pair with a friend as study buddies, each of you can see the other's weekly lesson counts and your shared streak while you are paired, and those weekly counts are kept as your pair's history after it ends. A friend pairing needs both friends to agree, and either of you can end it at any time. Study buddies can send each other short fixed messages from a list (like "Nice work!"). There is no free text, and the messages are kept as your pair's history. If you want a study buddy who is not a friend, you can choose to be matched with another learner of the same course at a similar level (you confirm you are 13 or older first). A matched learner sees only your display name, your avatar, your weekly lesson counts, your shared streak, the preset messages you send, and that you study the same course at a similar level. You can end the pairing, block or report them at any time, and you can stop looking for a match whenever you like. If we pause matching, matched study buddies cannot send messages until it resumes. Friend pairings are not affected.

You can change your display name in the iOS app under Profile → Settings, or on the website on your Profile page. Everything else, including your email address, your answers, your recordings and your conversations, is never shown to other users.

We use the following processors. Each acts on our instructions and only for the purpose listed.

• Supabase: database, authentication and server functions. Holds your account and all learning data.
• Vercel: hosts the website and the server API that the apps call. It handles every request the apps and the website make, and keeps request logs (including IP address and device or browser type) for a limited period so we can keep the service secure and working.
• Resend: delivers the emails we send, such as your sign-in codes and password-reset links, and internal alerts to us when a learner files a report.
• Deepgram: converts your recorded speech to text, AI replies to spoken audio, and our audio library scripts to the AI-narrated episodes. It receives the recording or the reply text, not your account details. Deepgram does not keep your recordings or use them to train its models: we opt every request out of its model improvement program, so it keeps the data only as long as it needs to process the request.
• OpenRouter: passes the text below to the AI model that answers it and returns the reply. It sends the text only to the AI providers named in this list, and only to providers that do not train on it. It does not keep the text, apart from anonymous usage statistics.
• Anthropic (Claude), reached through OpenRouter: generates AI replies for Hector and the practice scenarios, grades written translation answers, writes the meaning and answer choices when you save a word, explains a mistake, a word or a message in your own language when you ask, and finds practice items in your conversation mistakes. It receives only the text needed for each task. Anthropic does not use it to train its models and deletes it within 30 days, or keeps it longer only when it is flagged for breaking Anthropic's usage policy.
• RevenueCat: manages subscription status, so Alphonso Pro works on every device you sign in on. It receives an account identifier and purchase and renewal information from Apple, Google or Stripe, not your payment details. For a subscription bought on our website it also runs the checkout page.
• Stripe: processes card payments for subscriptions bought on our website. It receives your card details and billing email directly; we never see or store your card number.
• Sentry: receives the crash and error reports described above, so we can find and fix faults. It stores the technical details of the fault for 90 days, not your account details, and does not use them for advertising.
• PostHog (United States data region): receives the anonymous usage data described above, only if you have opted in, so we can see which features are used and improve lessons. It does not receive your name, email or account identifier, and does not use the data for advertising.
• Apple: Sign in with Apple if you choose it, App Store billing for Alphonso Pro, and the Apple Push Notification service for notifications on iPhone.
• Google: Sign in with Google if you choose it and, on Android, Google Play billing and Firebase Cloud Messaging for notifications.
• Neon: hosts the database behind the launch email list on our marketing website. It stores the email address you enter there and when you entered it.
• Vercel Web Analytics: counts page views and a few events, such as joining the launch email list, on our marketing website. It counts them in aggregate and does not use cookies.

Our marketing website (discover.alphonsoecosystem.app) has no accounts. If you join the launch email list there, we store your email address and the time you joined in a database hosted by Neon, and we use it only to tell you when the iPhone app is available. Write to support@alphonsoecosystem.app and we will delete it. The site uses Vercel Web Analytics, which counts page views and events such as joining the list, in aggregate and without cookies.

Several of these processors are based in the United States, and some may process data in other countries. Where your data is transferred outside your own country, we rely on the transfer safeguards those providers make available, such as standard contractual clauses.

Hector, our advanced AI voice tutor, runs on the same account and the same systems as the rest of Learn with Alphonso. There is no separate sign-in and no separate account. Your spoken audio is transcribed by Deepgram and Hector's replies are generated by Anthropic's Claude through OpenRouter, as described above.

We do not keep the conversation itself on our servers after Hector answers. When a conversation ends, the app sends it to the AI provider once more to look for mistakes you make often. If it finds any, we save practice items to your review queue and a record of the type of mistake. That is how Hector remembers your weak spots between sessions. The same happens after practice scenarios and campaigns.

Because Hector is part of your one account, deleting your account removes your Hector data along with everything else. There is nothing separate to manage.

The app grades answers automatically, uses AI to find practice items in your conversation mistakes (only if you have allowed AI features), and uses your answer history to decide which items to review and when. A placement test estimates your starting level. These affect what the app shows you and nothing else. They have no legal or similarly significant effect on you, and a human is not involved. You can retake the placement test at any time.

When you report a learner, a team name or an AI reply, we store the report with the reason you chose and who or what it is about. For an AI reply, we also store the text of that reply and where it appeared. We are alerted by email and review every report within 24 hours. The learner you report is not told who reported them. When you block someone, we store the block so that it works on every device.

Reports, and the actions we take on them, are kept while the accounts involved exist, and for up to 90 days after one of them is deleted, so we can deal with repeated abuse. Deleting your account deletes the blocks you set.

We keep your account data for as long as your account exists. When you delete your account, your data is removed immediately from our live systems and is purged from routine backups within 30 days. Reports about content are kept for up to 90 days after an account is deleted, for safety, without any personal profile data. Request logs held by Vercel are kept only for the limited period its service retains them.

Data is encrypted in transit. Access to the database is restricted by row-level security so that one account cannot read another's records, and administrative access is limited to the people who need it. No system is perfectly secure, but we design on the basis that a leaked client key must not be enough to reach anyone else's data.

If you are in the UK, the EEA or a region with similar law, you may access, correct, export, restrict or delete your data, object to processing, and withdraw consent where we rely on it.

• Export: on the web, Profile → Your data → Download my data; in the app, Profile → Settings → Account → Export My Data. Both give you the same machine-readable JSON copy.
• Delete: on the web, Profile → Your data → Delete my account; in the app, Profile → Settings → Account → Delete My Account (type DELETE to confirm). Both erase your account and all associated records permanently.
• Correct: change your display name in the iOS app under Profile → Settings, or your display name and country on the website on your Profile page, or write to us and we will change them for you.
• Withdraw consent: turn off AI features in the iOS app or on the website as described above, and turn off notifications in your device settings.
• Anything else: write to support@alphonsoecosystem.app and we will respond within one month.

You also have the right to complain to your local data protection authority. In the UK that is the Information Commissioner's Office.

If you signed in to the iOS app with Apple, deleting your account also revokes that Sign in with Apple connection automatically. If that step ever fails on our end, you can revoke it yourself at any time, on your device under Settings → [your name] → Sign-In & Security → Apps Using Apple ID, or at appleid.apple.com.

We do not sell your personal information and we do not share it for cross-context behavioural advertising, so there is nothing to opt out of. You have the right to know what we collect, to request a copy, to request deletion, and not to be discriminated against for exercising those rights. The export and deletion tools above are open to everyone, wherever you live.

Learn with Alphonso is not intended for children under 13, or under the higher age that applies where local law requires one. We do not knowingly create accounts for them, and study buddy matching asks you to confirm you are 13 or older. The app is rated 13+ on the App Store.

When you start, Alphonso asks your age range. If you choose Under 13, the website stops there: no account is created, nothing you answered is sent to us, the answers are wiped from your browser, and your browser keeps a note of that choice so sign-up stays closed on that device. We store an age range, never your date of birth.

Every AI reply follows safety rules written for learners aged 13 and up, is checked against a list of blocked terms before you see it, and can be reported from the reply itself. We follow our AI provider's requirements for apps used by teenagers and the child-privacy laws that apply to us.

If you believe a child has created an account, write to support@alphonsoecosystem.app and we will delete it.

The website stores a small number of items in your browser, all listed in our Cookie Policy. The apps do not use cookies. They store your sign-in session on the device so you stay signed in.

We will update this page when our practices change and revise the date above. Significant changes will be highlighted in the app.