Ask where the number came from
Before interpreting a number, identify its source. Did you enter it? Did a connected device supply it? Was it calculated from other inputs? Was it generated from a photo or a description? Those questions change how much confidence the result deserves.
Portions matter
A meal estimate is sensitive to portions and ingredients. A photograph may not reveal cooking oil, a hidden ingredient or the weight of a serving. Reviewing the proposed values is part of using the tool, not an optional extra.
Different sources can disagree
A workout calorie estimate and a wearable estimate may use different assumptions. Sleep records may use different grouping and time boundaries. Comparing values without matching source, date, units and method can create a difference that is difficult to interpret.
Use uncertainty as information
An unavailable result can be a useful signal that the app needs more evidence. It is better to keep that uncertainty visible than to treat missing data as a reassuring zero. An estimate helps organize a question; it does not automatically answer a medical one.