AI Processing Notice

Last updated: 1 October 2026

Why this notice exists

Most of Tellsy works entirely on your phone. AI features are the exception: to write an AI read, a short part of your chat has to be sent to an AI model. This notice explains exactly what is sent, where it goes and what happens to it.

Which features use AI

• Full AI report
• Ask Tellsy
• Decode
• Reply coach
• Love language signals, Closure report, Red flag deep scan, Compatibility deep dive
• AI chapter names and "What changed" comparisons

These features need your separate, explicit consent. Free features (stats, the Wrapped story, the Chat Aura, texting types, "Who's more likely") never use AI and never send message content.

What is sent

For each AI request, Tellsy prepares on your phone:

• A short excerpt of the chat, not the whole chat. It is selected from recent messages and a few sampled moments (for example the start of the chat, busy days and quiet periods), up to a fixed size.
• Statistics calculated on your phone, as numbers only (for example message counts, reply times and active hours).
• For Ask Tellsy and Reply coach, the question or draft you type.
• The language you use Tellsy in, and the ID of the feature.

How it is masked

Before the excerpt leaves your phone:

• your name becomes "Me" and the other person's name becomes "Them", including common short forms of those names inside messages;
• phone numbers, email addresses and links are replaced with placeholders such as [phone], [email] and [link];
• long number sequences that look like IBAN, card or national ID numbers are replaced with [number];
• photos, voice notes and stickers are never sent; they appear only as [photo], [voice] or [sticker].

Limits of masking: names of other people mentioned in messages, addresses written in words, and other personal details may not be detected and may be sent as written. You can check the exact text of every request in Privacy center > "See what was sent to AI".

Where it goes

1. Your phone sends the masked excerpt over an encrypted connection (TLS) to our server.
2. Our server checks that you are signed in and have access to the feature, then forwards the request to our AI provider: a third-party large language model service whose servers are in the People's Republic of China.
3. The provider generates the result and returns it to our server, which passes it back to your phone.

What is kept, and where

• On our server: the excerpt and the result are not stored. They are held in memory only while the request is processed. We log technical information only: time, feature, token count, duration and error codes.
• AI answers you report: if you tap Report on an AI answer, the answer text shown on your screen (which can include first names) and your optional note are sent to us. We keep them for 90 days to review the answer, then delete them. They are also deleted when you delete your account.
• At the AI provider: under its own terms, the provider may keep request data and use it to improve its services. We cannot shorten or delete what it keeps, which is why only a short, masked excerpt is ever sent.
• On your phone: the result is saved with the chat, encrypted. A log of what was sent is kept on your phone only, and you can clear it at any time.

We do not use your chats to train AI models. The provider's terms allow it to use request data to improve its own services, so please do not use AI features on anything you would not want shared in masked form.

International transfer

The AI provider may process your excerpt outside Türkiye and outside the EEA/UK (in the People's Republic of China, which has no adequacy decision). This is covered by your explicit consent and by the safeguards described in our Privacy Policy and KVKK Notice.

What AI results are, and are not

• AI results are reads of patterns in text: word choice, timing, who starts conversations, how quickly people reply.
• They are for reflection and entertainment. They can be wrong, and they cannot know tone of voice, context outside the excerpt, or what happens offline.
• They are not professional advice and not a verdict on anyone. They do not detect lies or cheating.
• Every claim in a Tellsy AI read links to a quote from the chat. Tellsy checks on your phone that each quote really appears in the chat and removes any it cannot find.
• If an excerpt suggests that someone may be in danger, Tellsy shows support resources instead of a relationship read.

Your choices

• You can withdraw consent at any time in Privacy center > AI features. This stops new AI requests. Results you already have stay on your phone until you delete them.
• You can delete any chat and its results, or all data, at any time.
• You can see and clear the AI log on your phone.

Contact

MOZ GROUP TEKNOLOJİ BİLİŞİM SANAYİ VE DIŞ TİCARET LİMİTED ŞİRKETİ Halkalı Merkez Mah. 1. Posta Sk. Cadde24 Rezidans No: 12 İç Kapı No: 77, Küçükçekmece / İstanbul, Türkiye privacy@tellsy.neuros.com.tr