Privacy Policy

Last updated: 5 September 2026

This page is an English summary for partners and for anyone reading in English. The binding, complete version is the Polish one: Polityka prywatności. Where the two differ, the Polish text governs.

How we handle your photos

Selfies you upload for colour analysis are processed in memory only and discarded immediately after analysis. We never save them to disk or keep them on our servers — only the derived colour values (your season and palette) are kept. This zero-retention rule applies only to biometric photos (selfies, and body photos for fit analysis).

Clothing photos are not covered by that rule — they are ordinary product content we're entitled to keep, since that's what the app is for. This also applies to photos where someone else is visible (e.g. a mirror reflection while trying something on): we treat them like any other clothing photo. Today, as a matter of current implementation (not a privacy commitment), we keep a small thumbnail of each garment so your wardrobe can display it, and we don't currently keep the full-resolution original. The boundary that actually protects privacy is that we never run face processing on clothing photos — no detection, no skin-tone extraction, no linking to your colour profile — regardless of who is visible in the shot. Deleting a scan from your history removes the entry together with its photo; removing a garment from your wardrobe also deletes its photo, while the entry itself is only hidden (saved outfits and the calendar still refer to it). If the same photo is still used somewhere else, the file is removed once you delete the last place using it. Deleting your account removes everything. Your consent as the account holder cannot cover anyone else, so our terms require you not to upload photos of other people without their knowledge and consent — a contractual limit on you, not a technical safeguard, which is why we additionally never process faces on clothing photos. See the Polish deletion guide.

Biometric colour analysis is opt-in

Analysing a selfie to determine your colour season uses facial colour information. This only ever runs with your explicit consent, which you can withdraw at any time. You can also choose your season manually and skip facial analysis entirely.

Signing up as a tester

Flattera is in closed testing before its Google Play release. If you sign up as a tester, we store the email address and the details you enter on that form, and we use them only to run the test — inviting you to the closed track and contacting you about it. We do not sell your address or use it for anything else. The tester sign-up form is in Polish: Zostań testerem.

Partnership enquiries

If you write to us through the partnership form, we store the company name, the email address, the message and — optionally — the contact person and the affiliate network you came from. We use them solely to answer the enquiry and, where relevant, agree a partnership; the basis is your consent (Art. 6(1)(a) GDPR) and steps taken prior to entering into a contract (Art. 6(1)(b) GDPR). The enquiry is stored in a Cloudflare database (our hosting and database provider) and deleted no later than two years after our last contact, or earlier at your request.

AI processing

Some analysis is performed by a third-party AI provider. Selfies (with your consent) and clothing photos go to Google Cloud Vertex AI where they are processed inside the European Union. A fallback path to Anthropic (Claude) exists in the code but is inactive today: production holds no API key for it (verified 25 August 2026), so in practice photos go to Google only.

Your photos are never used to train the provider's models — with Google this is a contractual commitment that applies regardless of our settings. We do not keep the photo after we receive the response.

Zero data retention at Google — arranged on 1 September 2026. Not training on your data is not the same as not storing it at all, so we spell this out separately. Zero data retention is not the default in Google Cloud; it has to be arranged for the specific project, and we did that: on 1 September 2026 Google approved an exception for our project, exempting our requests from prompt logging for abuse monitoring. From that date Google does not store the photos we send — neither for model training (contractually guaranteed all along) nor for investigating breaches of its own policy.

How it worked before. We are keeping this description because it covers photos sent before 1 September 2026. Google's standard mechanism applied then: if its automated safety classifiers flagged a request as suspicious, Google could store its content — that is, the photo — for up to 90 days, solely to check whether its acceptable use policy was breached. The data was not used to train models, and any such record stayed inside the European Union — in the same location we selected for our project. Within that window an authorised Google employee could review the stored photo.

Where processing happens — verified 25 August 2026. Image analysis takes place in the European Union: we use a location for which Google keeps processing of customer data inside EU member states. The United Kingdom and Switzerland fall outside that scope, and your data does not go there. We call this out separately because until 25 August 2026 this page named a specific city — Frankfurt — and that promise was not backed up: the model we use is not served in any single region, so requests were quietly handled by the global endpoint with no location guarantee. We changed the configuration, which is why we now say "in the EU" rather than "in Frankfurt": a narrower claim than the old one, but unlike it, true and measured. The Polish version of this policy is the authoritative one.

Cookies

This marketing site uses only what's needed to function. We do not run advertising trackers on it. If we add analytics later, we'll update this page first.

Contact

Questions about your data? Email hello@flattera.app and we'll help.