Privacy

The short version

• What you upload is yours. It is never used to build anyone else's plan, never pooled, never used to train a model.
• We do not keep the files you upload. We extract the text, use it to build your plan, and discard the file itself.
• Your goal, your material and your questions are sent to the model provider that writes your lessons. That is the product; there is no version of it that does not do this.
• If you arrived from an advertisement, the advertising network is told that you arrived, that you signed up, and what you paid. That is measurement of the advertisement, and it uses a cookie set by them.
• Our own analytics is a separate thing and is off until you turn it on — twenty-three named events, no cookie, and never anything you wrote.
• You can export everything we hold, and delete the account and all of it, from your account settings.

Who we are

howdoistudy is operated by Ernest Laptiev, Heorhiivska 42, Kyiv, Ukraine, 08140. For anything about your data, write to contact@howdoistudy.com. Under the UK and EU GDPR we are the controller of the data described here.

What we store

• Your account — your email address and the name you gave. If you signed up with a password we store a bcrypt hash of it and never the password. If you signed in with Google we store the link to that Google account instead, and there is no password.
• What you tell us you want to learn — the goal, where you are starting from, how much time you have, any deadline, and the whole conversation that settled your plan.
• Material you upload — documents, spreadsheets, slides, transcripts and images. We extract the text and store it in passages so your lessons can cite them. PDFs and images are read by our model provider, which means the file itself is sent to them to be read; nothing else you upload is. We keep the filename. We do not keep the original file.
• What you do in a lesson — which exercises you answered, what you answered, whether it was right, and which misconception a wrong answer pointed at. That is what decides which lessons come back to you and when; a learning tool that does not record this cannot adapt.
• Questions you ask — the conversation inside each lesson, and any passage you highlighted when asking.
• Operating records — sign-in sessions, and what each generation cost us in model tokens.
• If you turned analytics on — a row for each of the twenty events below, with the lesson or plan it happened in, how far through you were and how long it took. Nothing you typed. If you never turned it on, there are none of these.

Who else sees it

Four, and no others. We do not sell data.

• The model provider that writes your lessons — Anthropic, or Anthropic models operated by Amazon Web Services, depending on how this deployment is configured. Your goal, the material you uploaded, and your questions are sent there to produce the plan and the lessons. Anthropic’s commercial terms do not permit training on data submitted through the API.
• Amazon Web Services — hosting, the database, and the email that carries password reset links. Region: Ukraine.
• Google — if you choose to sign in with Google, to confirm who you are; and, where this deployment advertises, to measure what an advertisement led to.
• TikTok — only for measuring advertising, and only where this deployment advertises. Their pixel runs on every page and tells them which pages you visited. While you are signed in it is also given two identifiers for you: your email address, and a number standing for your account. Both are hashed before they leave this server — the account one with a secret only this deployment holds — so they can recognise somebody they already know and cannot read somebody they do not, and the account identifier means nothing anywhere else. When you pay, we tell them the amount and the currency directly from our server as well. Never what you are studying. They are a separate controller for what they do with it, under their own privacy policy.

None of it includes what you uploaded, what you asked, or what howdoistudy wrote for you. Advertising measurement is told that a payment happened and what it was worth; it is never told what you are studying.

What we never do with your uploads

If you upload a textbook, a syllabus or your own notes, that material builds your plan and nobody else’s. It is scoped to your account in the database, not by convention. The structure howdoistudy derives while reading it — the names of concepts, which one has to be learned before which — is ours and may be reused to make other plans cheaper to build. The material itself is not.

Cookies and what is kept in your browser

One that keeps you signed in. It is strictly necessary — without it there is no way to know the plans are yours — so it does not need your consent and we do not pretend to ask for it.

Where this deployment advertises, there are others, and they are not necessary, so they wait for the cookie banner. Until you answer it, and if you say no, Google’s tag sends only a cookieless ping that identifies no browser, TikTok’s pixel is held and sends nothing, and the click identifier an advertisement puts in the link you arrived on is kept only for the length of your visit. If you say yes, Google’s tag and TikTok’s pixel each set their own cookie to recognise this browser across visits, and ours holding that click identifier is kept for 28 days, so that a purchase days later can still be matched to the advertisement that led to it. Clearing your cookies removes all of them, and a browser or extension that blocks trackers stops the tags outright — nothing about howdoistudy works differently if you do.

Your browser also remembers your theme and whether the plan rail is folded. That never leaves your device.

There is analytics, and it is off unless you turn it on in your account settings. It uses no cookie and nothing in your browser: the record is a row in our database tied to your account, which is why turning it off actually stops it rather than merely stopping one copy of it.

If you do turn it on, what gets recorded is twenty things happening — a lesson opened, a lesson left before the end (with how far in you were and how long you had been there), a question asked, a widget shown in answer to one, a quick question asked outside a lesson (and the offer to answer one, and the week's allowance being met), a plan edited, a block rated, a lesson you were asked to pay for, and the ten steps on the way in from the landing page to a payment (signing up, seeing the plans screen, starting and finishing a first lesson, opening a checkout, paying), each with which version of those screens you were shown — and nothing you wrote. Not the question, not the edit, not the answer. Those are already stored above because the product cannot work without them; measuring is a separate purpose and it gets the count, not the text.

You can turn it back off in the same place, it takes one click either way, and nothing about the app behaves differently for someone who says no.

Separately, and about nobody in particular: we count how many times each page was opened, per hour. That is a number going up — there is no row for your visit, no identifier, no cookie, no address and nothing about your browser, and there is nowhere in it to put any of those. It cannot tell us that two openings were the same person, where you came from, or that the person who read this page is the one who signed up. It exists so that “how many people looked” has an answer at all, and because it is about nobody it needs no permission from anybody. The same kind of counter, with the same nothing in it, counts how many times each step on the way in was taken — a sign-up, a first lesson, a payment — so that we can see where people stop without knowing who stopped.

On the pages you can reach without signing in — this one, the landing page, the subject pages and the sign-in screens — we also use PostHog to see which parts of a page people actually click. It never runs once you are signed in: your lessons, your answers and your plans are not sent to anybody else, ever.

By default it keeps nothing on your device beyond the visit, which is why it ran for a long time with no banner: every page load was a new stranger to it. The banner asks one question that covers everything above which would keep something on your device — remembering this browser, so that somebody returning is not counted as somebody new, and the advertising cookies described in the section before this one. Saying no leaves the site identical and the click counts running exactly as described in the paragraph above; saying yes adds an identifier in your browser’s own storage, so that a return visit is counted as one. Until you answer, and after a yes, the public pages are also recorded for replay and heatmaps — with every field you type into masked, anything you wrote, uploaded or were shown from your own material blanked out, and your network address not sent — so a replay shows where someone went on the page and never what they said or who they are. A no stops that recording on the spot. Your answer is kept in a cookie, because a site that forgets you said no is a site that has not taken the answer.

We are also testing five versions of the front page. Which one you see is decided once, at random, and kept in a cookie for the length of your visit so it does not change while you are here; if you say yes at the banner that cookie is kept for a year, so you see the same version when you come back. It holds a random identifier and the name of the version, and nothing about you.

There is no replay on any page behind sign-in, ever: your lessons, your answers and your plans are not recorded by anybody. If you said yes at the banner and then made an account, our server tells PostHog about a few milestones after sign-in — a plan built, a first lesson started and finished, a checkout opened, a payment made — under a scrambled identifier, so that we can tell which version of the front page led to them. Never what the plan is about, never a word you wrote.

How long we keep it

Until you delete it. Deleting your account removes your plans, lessons, answers, progress, conversations and sign-in methods immediately — it is a real delete, not a flag. Backups roll off within 7 days.

Your rights

You can get a copy of everything we hold as a single file, and you can delete the account and all of it, from your account settings — no email, no waiting. You can also object to processing, ask us to correct something, or complain to your data protection authority (in the UK, the ICO; in the EU, your national authority).

Age

howdoistudy is built for students, and plenty of students are children. You need to be 13 or older to use it. If you are under 16, check with a parent or guardian first — in several countries they have to agree on your behalf — and they should read this page too.

We do not show advertising on howdoistudy itself, and we do not use anyone’s work to train a model — those hold whatever your age. We do buy advertising elsewhere, and measuring it means the network that carried it is told you arrived and what you paid, as described above.

Changes

This notice carries a version. When it changes in substance we ask you to accept the new one before you carry on — we do not treat consent to an older version as consent to this one. See also the terms.