The short version: your screenshots are read once and destroyed within seconds. We keep the decode — never the pictures, never a transcript, never anyone's face or name from inside a chat.
When you run a decode, Luona receives the chat screenshots or pasted text you chose, at your instruction, after you agreed to it on a screen that explains what happens next.
We do not perform facial recognition on anything. Profile pictures that happen to appear inside a screenshot are not analyzed, extracted, matched or stored. We do not create face templates or identify anyone. The analysis reads the words, the pattern and the timing of a conversation to produce texting guidance.
Your upload is transmitted to Google, whose Gemini model performs the analysis on our instruction, and to no one else. Google processes it to return the result and, under its paid API terms, does not use it to train its models. We do not disclose your conversations to any other third party, and we will not do so without your separate consent, except where a court order or law requires it. We never contact anyone who appears in your conversations.
Before your first decode, the app shows you a screen describing exactly what is collected, why, and when it is destroyed, and asks you to agree. We record which version of that wording you agreed to and when. You may withdraw your consent at any time in Settings.
Write to hello@luona.app. See also our Privacy Policy and Terms of Use.