Privacy Policy
In short
Your library — saved videos, posts, finds and categories — is stored only on your iPhone and never reaches our servers. There is no sign-up, no ads, and no advertising identifier is read. Data leaves your device only for the specific feature that needs it. The AI features and category training work only after a separate consent that you can withdraw in Settings.
1. Who we are
The data controller is the independent developer of the Sohra app. Contact: hello@leanbe.online.
2. What stays on your device
The library is stored in the app's storage on your iPhone. Everything below stays there and is not accessible to us:
- saved videos, images and post texts;
- categories, subcategories and tags you created;
- speech transcripts, on-screen text and finds;
- viewing history, favourites, trash;
- tasks, reminders and film marks.
Speech and on-screen text are recognised on the device using iOS frameworks. Audio and video do not leave the phone. You create library backups yourself (Settings → Backup) and decide where to keep them, e.g. iCloud Drive or a computer.
3. What leaves your device, and when
| When | What is sent | Where |
|---|---|---|
| You save a link from another app | The link itself. No platform account or cookies. | Our server (api.leanbe.online), which fetches the item and its description and returns them to the app |
| AI features — only after your consent (Settings → AI → Cloud AI) | The item's text: caption, speech transcript, on-screen text — to suggest a title, a category and finds. Single video frames — only when you tap the picture analysis button. A random install number goes with the request to enforce the limit of 5 requests per hour. | Our server, which forwards the request to a language model by Groq Inc. (USA) |
| You turned on “Train auto-categories” (off by default) | The beginning of the item's text, caption, hashtags and the category you chose; link and author only for public platforms. The install number only as a hash that cannot identify you. | Our server, to train category sorting to work better without the AI |
| Films tab | The film title, to find a poster and rating | Our server, which queries a film database |
| Song recognition | An acoustic fingerprint of a few seconds of audio — not the recording | Apple (ShazamKit) |
| You open a product from a find | The product's article number or name, to open its page | The marketplace, directly from your device |
| Technical statistics | Anonymous events: app launch, our notifications shown and opened, app and iOS version, random install number | Our statistics server |
The random install number is created on first launch. It is not an advertising identifier, is not linked to your Apple ID, phone or e-mail, and changes when you reinstall the app.
4. What we do not collect
- name, phone number, e-mail — there is no sign-up;
- the advertising identifier (IDFA) — there are no ads and no tracking, so no tracking prompt is shown;
- location, contacts, calendar, messages;
- your library as a whole — only the fragments listed above, for a specific feature;
- payment data — there are no in-app purchases.
5. Purposes and legal bases
| Purpose | Legal basis |
|---|---|
| Saving an item from your link | Performance of a contract — you ask for it |
| AI titles, categories and finds | Your consent, withdrawable in Settings → AI |
| Training auto-categories | Your consent, withdrawable in the same place |
| Technical statistics, request limits, abuse prevention | Legitimate interest in a stable service. Data is anonymous |
6. Third parties
- Groq Inc. (USA) — runs language-model requests. Receives the request text without information about you. Under Groq's terms, request data is not used to train models.
- Apple Inc. — app distribution, song recognition (ShazamKit) and iCloud Drive if you keep a backup there.
- A film database — posters and ratings; receives only the title.
- Marketplaces — only when you open a product yourself.
We share data only with parties that protect it at least as well as described in this policy, and only as much as a specific request needs. We do not sell data or share it with ad networks or data brokers.
7. Retention
- Library — on your device until you delete it.
- AI request texts — not stored by us after the response.
- Category training samples — up to 24 months.
- Anonymous statistics — up to 24 months.
- Server logs — up to 30 days.
8. Your rights
Consent to the AI features and to category training can be withdrawn in Settings; the feature stops sending data immediately. Deleting the app erases the library on the device; delete any backups you made yourself. If you are in the EEA or the UK you have the rights of access, rectification, erasure, restriction and portability, and the right to complain to a supervisory authority. Statistics and samples are anonymous, so we cannot find your records by name or e-mail — write to us and we will see what we can do.
9. Children
The app is not directed at children under 13 and does not knowingly collect their data.
10. Security
Traffic to our servers uses TLS. Keys to external services are kept on the server and are never shipped in the app. Admin panels are password-protected.
11. Changes
We announce changes on this page and update the date above. Material changes are also shown in the app.