Review this draft
Nothing is logged until you confirm. Every item is editable, and the quantities are the part most worth checking — identification is the reliable half of this, portion estimation is not.
The model decided
Which foods are in the photo, roughly how much of each, how it was prepared, and how sure it is. That is the whole list.
The database decided
Every nutrient value below. The model was never asked how much protein or magnesium anything contains — so a misidentification is visible and correctable rather than an invisible wrong number.
What the model noticed about the photo: The left edge of the plate is out of frame, so the rice portion is likely underestimated. Warm overhead lighting makes the greens read darker than they are.
2 items are low confidence. Check them before confirming.
Judged size by: roughly deck-of-cards sized, judged against the fork beside it
Matched in the food database to
Judged size by: about one cup, judged against the rim of the bowl
Matched in the food database to
Not right? Rice, brown, long-grain, cooked · Rice, jasmine, cooked
Judged size by: small side portion, partly hidden behind the chicken
Matched in the food database to
Not right? Kale, boiled, drained · Chard, swiss, cooked
Judged size by: visible sheen on the vegetables only — the amount is a guess
Matched in the food database to
Not right? Butter, salted · Oil, canola
What this adds to your day
Every figure here comes from the resolved food's row in the database, scaled by the quantity above.
Energy
593 kcal
Protein
54.3 g
Magnesium
152 mg
Vitamin C
9.8 mg
Confirming is the only path from a draft to your log. There is deliberately no auto-accept, no matter how confident the model is — an AI suggestion is a draft until a person says otherwise.