Legal

AI Privacy Notice (Charlie & Photo Scan)

Last updated: June 9, 2026

This notice explains how NatBuddy's AI features handle your information. It supplements our Privacy Policy, which governs all of your data.

1. Why we publish this notice

Some NatBuddy features use artificial intelligence to generate responses and analyze photos. Because AI features involve sending certain information to an outside AI provider, we want to be clear and specific about what is sent, what stays on your device, how it is handled, and how you stay in control. The AI features are off until you turn them on.

2. What Charlie is

Charlie is NatBuddy's AI wellness companion and brand voice. It answers questions about dog nutrition, recipes, portions, and product labels in plain language, drawing on peer-reviewed veterinary nutrition literature. Charlie is a wellness companion, not a symptom checker. It does not diagnose, and its responses can be incorrect.

Charlie is an AI companion. Its responses are informational, can be incorrect, and are not veterinary advice, diagnosis, or treatment. Always consult your veterinarian.

3. What is sent to our AI provider

When you use an AI feature with consent enabled, the relevant input is sent through our secure server (a Supabase Edge Function) to our AI provider, Anthropic, which generates the response. Depending on the feature, this can include:

Our AI provider's key is held only on our server and is never shipped inside the app. Every AI request is sent over an encrypted connection.

4. What stays out and what stays on your device

We do not send your email address, password, or payment information to the AI provider — those are never part of an AI request. If you dictate a message to Charlie by voice, the speech is transcribed to text on your device using Apple's on-device speech feature; the audio is not uploaded, and only the resulting text leaves your device if you send it. We send the minimum pet context needed to answer a question, not your full account.

5. Retention and our no-training commitment

The information you send through NatBuddy's AI features is used only to generate the response you asked for. These AI inputs are not used to train AI models and are retained per Anthropic's commercial API terms. We do not sell AI inputs or use them for advertising. We retain your own NatBuddy data — such as your scan history and chat history inside the app — as described in our Privacy Policy, and you can delete your account in-app at any time.

6. Security of the AI flow

AI requests are routed through our server rather than directly from the app, so the provider key never leaves our infrastructure. Requests require a valid signed-in session, are rate-limited per user, and are logged for abuse prevention and cost control. Data in transit is protected with encryption (TLS), and access to your stored data is restricted by per-user database security rules.

7. Your control over AI features

AI features are opt-in. The first time you try to use Charlie or an AI-powered scan, NatBuddy asks for your explicit consent, and the features stay disabled until you grant it. You can turn AI features off again at any time in the app's settings. When consent is off, the server refuses AI requests, so no chat text or photo is sent to the AI provider.

8. Cautious language and limits

NatBuddy's AI features use cautious, non-alarmist language. The scanner avoids absolute terms like "bad" or "dangerous," reserving clear warnings for genuine toxics such as xylitol, and otherwise uses phrasing like "generally suitable," "some ingredients to consider," or "consider alternatives." AI responses and photo scans are informational, can be wrong, and are not a diagnosis or a substitute for your veterinarian. For any concern about your dog's health, consult your veterinarian.

Contact

Questions about NatBuddy's AI features can be sent to support@natbuddy.com, or by mail to Granviex, LLC, 4700 NW Boca Raton Blvd #202, Boca Raton, FL 33431.