Privacy Policy
GrammarLens has no accounts, never asks for your email, and shows no ads. Your name, your avatar and your entire practice history — including the Daily Test — stay on your phone; the Daily Test is scored on the device and never leaves it. If you turn on AI feedback for practice sessions (Topic Practice and practicing a weak spot), the answer you type and the question it responds to are sent through our server to Anthropic (Claude) so it can check your work — only after you've agreed to it in the app, and you can withdraw that agreement at any time.
Contents
Who is responsible
GrammarLens is made by Ahmet Emin Tayfur, Istanbul, Turkey. For anything in this policy, write to support@ahmettayfur.com.
What stays on your device
Everything below is stored in the app's local database and is removed when you delete the app. None of the content below leaves your device; the only exceptions are the settings and totals described at the end of this section:
- Your first name, from onboarding
- Your chosen avatar and your app settings, including text size and theme
- Your weak-spot profile — which grammar topics you get wrong and how often, worked out entirely on your device
- Every Daily Test question and answer — the Daily Test is scored on your device and never leaves it
Your practice history (Topic Practice and weak-spot practice) — the questions you're given and the answers you type — is also stored on your device. The one exception: if you've turned on AI feedback, the answers and matching questions from the session you submit are sent off-device for scoring; see Permission for AI feedback for exactly what that involves and how to control it.
Separately, the learning goal you choose during onboarding, your text-size preference and totals about your sessions — like how many questions you answered or skipped — are sent to our analytics provider as numbers and category labels, never as text; see Analytics and crash reporting.
The Daily Test
The very first day's test is built into the app itself, so it needs no network request at all. On later days, the app asks our server for that day's set. The request carries nothing about you or your history. Every Daily Test, on any day, is scored entirely on your device.
Permission for AI feedback (practice sessions)
Practice sessions — Topic Practice and practicing a weak spot ("Practice this") — are where GrammarLens checks what you write and explains your mistakes. They are the only place your own words leave your device, and only once you say yes.
Before you use it for the first time, the app shows a permission screen that states what is sent (the answer you type, together with the question it responds to), who receives it (Anthropic, the company that makes Claude, reached through our own server), and what is never sent (your name, your learning goal or your avatar — Daily Test answers stay on your device regardless). You can decline and still use the Daily Test.
You can review or withdraw this permission at any time in Profile → Data. Turning it off takes effect immediately; the app will ask again the next time you start a practice session.
Please don't type personal details — such as your full name, address or contact information — into your answers, since we can't screen what you write before it's sent.
What our server keeps
Our server (a Cloudflare Worker) never stores the text of your questions or answers — it passes each request to Anthropic and returns the result. It keeps three things, briefly, as operational records only:
- Usage counters. A random, anonymous identifier — created by the app on your device, not your device's serial number, advertising ID, or any other hardware or system value — is used only to enforce a daily limit on how many exercises one installation can generate, which is what keeps the service affordable to run. Limits are scoped to the calendar day (UTC) and counters delete themselves automatically 48 hours after they are last written. This identifier is never sent to Anthropic. Deleting and reinstalling the app produces a new identifier.
- A request log, for every completed call. Which kind of operation it was, how many questions it covered, how many tokens were used, and how long it took — never the questions or answers themselves.
- Error logs. When something goes wrong — the AI service is unreachable, or returns something the app cannot read — our server records the failure so the problem can be found and fixed. These records do not contain the text you wrote either.
Both logs are kept only as long as our hosting provider, Cloudflare, retains Workers log output on our current (free) plan — about three days — and are then deleted.
Anthropic as a processor
When you use a practice session with permission on, the answer you typed and the matching question are processed by Anthropic to generate your feedback. How Anthropic retains that data, and whether it is used for training, is governed by Anthropic's own terms and policies, not ours to restate here. See Anthropic's Commercial Terms of Service and Anthropic's explanation of API data and training for the current terms.
Analytics and crash reporting
GrammarLens uses Google's Firebase Analytics and Firebase Crashlytics to understand how the app is used and to find crashes.
What we send deliberately falls into a few categories:
- Which screens and modes you open
- How practice sessions and the Daily Test go — like how many questions were answered or skipped
- Progress milestones, such as medals
- Steps through the paywall and purchase flow
- Your learning goal from onboarding — exam prep, work, everyday confidence, or skipped
- Your text-size preference
- Your AI-feedback permission decision (see Permission for AI feedback)
The learning goal you choose during onboarding is sent with the app's usage analytics, never with your name. It helps us decide what to improve; it does not change your lessons.
The exact events inside these categories may change as the app evolves; what doesn't change is the boundary — none of them ever carry your name or the text of any question or answer.
What Firebase collects automatically. Independently of the events above, Firebase records a randomly generated app instance identifier, your device model and operating system version, your app version, your approximate region and language, and session counts and durations. Crashlytics additionally records technical details about a crash when one happens — device and app information, the error message, and where in the app it occurred.
This is not linked to your identity. GrammarLens does not use an advertising identifier (IDFA), and we do not use this data to track you across other apps or websites. If you would rather not send this, iOS settings let you limit app tracking and diagnostics at the system level.
Subscriptions
GrammarLens Premium is sold through Apple's App Store. We never see your payment details — your card, your Apple ID and your billing address stay with Apple.
To know whether your subscription is active, the app uses RevenueCat, a subscription management service. RevenueCat receives an anonymous app user identifier, the purchase and renewal status of your subscription, and basic device and operating system information. It does not receive your name, your answers, or your practice history.
What we never send
Regardless of feature or permission state, GrammarLens never sends:
- Your name or your avatar
- Your stored practice history (other than the practice answers you choose to submit for feedback — see Permission for AI feedback)
- Daily Test questions or answers — the Daily Test is scored entirely on your device
- Your contacts, your precise location, or any advertising identifier
- Your payment details — Apple handles those directly
And more generally:
- No advertising, and no advertising identifiers
- No selling or sharing of personal data with anyone
- No tracking of you across other apps or websites
- No account, no password, no email address required to use the app
Keeping and deleting data
- Everything stored on your device is deleted when you delete the app.
- Usage counters on our server delete themselves after 48 hours.
- Request and error logs are kept for about three days by Cloudflare.
- We do not hold a copy of your practice history, so there is nothing for us to send you or delete on request.
- Analytics and crash data are held by Google under Firebase's own retention settings.
Children
GrammarLens is intended for users aged 13 or over, or the minimum age required in your country, whichever is higher. We do not knowingly collect personal data from anyone younger.
Your rights
Depending on where you live, you may have the right to access, correct, delete or export your personal data, to object to processing, or to complain to a supervisory authority. Because the app holds your practice data on your own device and we do not maintain an account for you, most of these rights you can exercise directly — deleting the app deletes the data. For anything else, write to support@ahmettayfur.com.
Changes
If this policy changes in a way that affects what we do with your data, we will update the date at the top of this page and, where the change is significant, note it in the app.