Trust isn't a feature.
It's the architecture.
Five rules we built LivFit on. Each is enforced in the product, not just stated here.
Never sold. Never used for ads. Delete any meal, any household member, or your entire account anytime. Stored encrypted at rest. Access logged.
All nutrition values come from USDA or FDC, branded databases, or your own kitchen. The LLM operates only on structured outputs.
Calorie and macro targets are never displayed to or generated for users under 18. Parents see patterns and explanations only. Enforced at the model layer via validate_compliance().
The AI may parse a receipt, scan a label, or draft a meal log — nothing reaches your record until you tap confirm. Enforced server-side via useAIDrafts().
Eating-disorder, child-prescription, and mental-health adjacent queries return one mode label and decline. The AI doesn't try to answer.
If we ever fall short of these rules, tell us at trust@livfit.ai and we'll fix it.