Short and human. The full legal text is approved by a lawyer.
The data controller is Oksana Anatoliivna Bodnar, a sole proprietor registered in Ukraine. Data and privacy questions: hello@tomorrowell.com, subject "Privacy".
What you log: sleep, energy, mood, food, notes — this is health data, and we process it only with your explicit consent, given as its own step at registration.
The app, the database and this site run on a netcup (Germany) server in an EU data centre. The database sits on an encrypted disk, and backups are encrypted before they leave the server. The encrypted backups are kept in Cloudflare R2 with Object Lock, which blocks changing or deleting them. Their location is set to the EU as a hint, and Cloudflare does not guarantee it. Backups older than 30 days are deleted. No ads, no selling data, no third-party trackers.
Processors the service cannot run without: netcup for hosting (Germany, EU data centre); Cloudflare R2 for encrypted backups; Paddle for paying for Plus and Max subscriptions (described below); Anthropic for the AI assistant and for translating dish names (described below), which may process data outside the EU; the USDA, Fineli and BLS nutrition references (queries to USDA carry no personal data, Fineli and BLS are kept in the app as tables); Zoho (ZeptoMail and Zoho Mail, EU data centre) for the app's emails and our mailbox: the link to confirm your address or set a new password, the reminder before a trial ends, and the emails you send us (Zoho receives your address and the text of the message).
Only when you choose the feature yourself: Google, if you sign in with Google; your browser's notification service, if you turn reminders on. Both are described below.
About the AI assistant, honestly: it runs on an Anthropic model, so the text of the messages you write to it is sent to Anthropic every time you use it — a technically unavoidable part of the feature. Your diary data is a separate matter: it joins the request only under the separate "personal answers" consent (off by default), and never as your whole diary. The AI assistant is optional: the diary and forecasts work without it.
Translating dish names: the nutrition references are in English, so a dish name written in a non-Latin script may be sent to Anthropic for translation into English. Only the name goes, with no link to your account. The translation is kept so the same name is not sent twice. The AI assistant does not receive food diary entries or photos. A meal comment on Plus and Max, under the "personal answers" consent, sends only what was found (for example caffeine or a late meal) and the time of the meal.
We do not keep food photos. There is one short exception, and only where the app can split a plate into dishes: when you save a dish recognised from a photo, that photo is held encrypted for at most 15 minutes so a background check can see whether the plate holds several dishes. As soon as the check is done (or after 15 minutes if it is not), the photo is deleted. What remains is only a suggestion to "split into separate entries", with dish names and each one's share of the plate — no photo and no outlines; unanswered, it disappears after 7 days. The purpose is the same: recognising the dish you photographed yourself.
Only if you turn it on: a separate consent, “Remember my dishes”, in Profile, off by default. Then, for each dish you confirm from a photo, we keep its numeric fingerprint — a set of numbers computed from the photo — together with the name you chose, the weight if you gave one, and the plate if you picked one. The photo itself is never kept. The fingerprint is compared only with your own records, so your past dishes help recognise the next photo, and it is shared with no one.
A fingerprint is deleted with the entry it came from and with your account. Turning the consent off erases every fingerprint at once. They are part of your data export.
Plates you add in Profile are a name and a diameter, used to estimate portions. They are also part of the export and are deleted with your account.
Only if you switch it on: a separate consent, "Help improve forecasts", in the Profile, off by default. Then the shared forecast model may learn from your logged days and how the following day went. Only those numbers go into training — no name, email, notes or photos, under a code instead of your account, with a day number instead of a date.
You can switch it off at any time: from the next training on, your data is not in it. Honestly about the limit: a version of the model already trained cannot unlearn — but the next one is trained without you. Your diary stays yours and does not go anywhere; forecasts work the same with or without this consent.
If you choose "Continue with Google", Google tells us your email address and the permanent identifier of your Google account. We do not ask for your name, photo or contacts. The identifier is kept so we recognise you next time; it is part of your data export and is deleted with your account.
The sign-in itself happens on Google's page and under Google's terms, so Google knows you are signing in to Tomorrowell. An account can be created through Google only after the same consent to process health data as a registration with a password. Signing in with a password works without Google.
Only if you turn them on: the "Remind me every day" button in Profile, and permission for notifications in your browser. Then we keep the address your browser issued for notifications, two encryption keys from the browser, the time you chose and your time zone. If the day is already logged, no reminder is sent.
The reminder is delivered by your browser's notification service: Google, Apple, Mozilla or Microsoft, depending on the browser. Its text is encrypted with your browser's keys, so the service delivers it without reading it. None of your entries are in it.
You can turn reminders off in Profile, and the subscription is deleted at once. If you block notifications in your browser's settings instead, we delete it at the first attempt to send. It is also part of your data export and is deleted with your account.
Plus and Max subscriptions are sold by Paddle as the Merchant of Record. When you pay, Paddle receives your name, email, country, billing address and payment details and keeps them; card details never reach us. Paddle processes this data as an independent controller under its own privacy policy, to perform the purchase contract and meet its tax obligations.
From Paddle we receive only what access to your plan needs: your email, customer and subscription identifiers, the plan, its status and the dates of the paid period. The basis is performance of the contract. Paddle receives none of your diary entries.
Paddle is a UK company and transfers data among its companies in the UK, the EU and other countries. For countries without an adequacy decision, Paddle uses standard contractual clauses. Requests about the data Paddle holds can also go to Paddle directly: privacy@paddle.com.
If you write to hello@tomorrowell.com (feedback, a question, a bug), we keep the email, your address and any attachments. The purpose is to reply and to fix or improve the app; the basis is legitimate interest. Emails are not added to your account and are not used to train models. We delete them once the matter is closed, and no later than 12 months after. Please do not send health details you do not want to share; screenshots may contain diary entries.
Export everything in one file and delete the account from Profile, any time, with no support tickets. Deleting asks for your password and a typed word to confirm. Your data is erased from the database at once, encrypted backups roll off within 30 days, and a record that an account was deleted stays, with nothing about you. The details are on the Your data page. Every consent switches on and off separately.