Features
Yes. Photograph the nutrition label and Sayved reads the per-serving or per-100 g figures, then scales them to the amount you ate. The numbers come from the pack, so the only real uncertainty left is how much you had.
A nutrition label is the one place on a food item where the numbers are already measured. Sayved uses that. Instead of guessing a portion from a picture, it reads the figures on the pack and works out your total from the amount you actually ate. If you have ever copied a label into a calculator by hand, this removes the copying.
You tap the floating button, choose photo, and take a picture of the nutrition table. Sayved sends a downscaled copy to an AI model, which reads the per-serving or per-100 g values, including energy, protein, carbohydrate and fat. It then multiplies those values by the amount you say you ate.
The amount can come from three places. You can say it while you capture (“I had half the tub”), you can answer a follow-up afterwards, or you can do neither and let Sayved assume one label serving. In that last case the entry is marked low confidence and carries a note that it assumed a single serving, so you can see exactly where the guess is.
This is a different problem from photographing a plate. On a plate, Sayved judges a portion by eye, which is why plate estimates come with wide ranges. On a label, the portion is the only unknown. Sayved treats the two differently, and you do not have to tell it which one you took. It recognises a label when it sees one.
Yes, and that is deliberate. A rule of the app is that capture is never blocked by a question. If you photograph a label and say nothing, the entry is saved immediately, one serving is assumed, and that assumption is recorded.
Only in that one situation does a single follow-up appear after saving: “how much did you eat?” You can answer with one tap, and Sayved re-reads the same stored photo with your answer and updates the entry in place. Same entry, same date, only the estimate changes. Plate photos never see this prompt, and neither do labels where you already stated the amount, so most captures end with the usual quiet confirmation and nothing else.
If you ignore the prompt, nothing breaks. The entry stays as an honest, low-confidence one-serving figure, and you can fix it any time from History.
Take a tub of cottage cheese whose label lists 98 kcal and 11 g protein per 100 g. You eat roughly 150 g and say so. Sayved scales the table by 1.5 and files about 147 kcal and about 16 g protein. The range stays tight because the label values are measured; it widens a little to cover your own estimate of the 150 g.
Now take a bag of crisps listing values per 30 g serving, and you say “I finished the bag”, which is 150 g. Sayved reads the serving size, works out that the bag is five servings, and multiplies. If you had only said “some crisps”, it would assume one 30 g serving and flag it, because it will not invent a quantity on your behalf.
The same logic suits supplements and protein powders. A photo of a tub label plus “one scoop” gives a clean entry. For daily doses you repeat, a saved routine is quicker; see the supplement tracker.
It is more grounded, which is not quite the same as exact. The figures come from the manufacturer and are usually rounded, and portion size is still on you. So a label entry typically has a narrower range than a plate entry, with a note about what remains uncertain.
For comparison, the calorie counter from photo page explains how plate estimates are built and why their ranges are wider. If a meal has both, such as a ready meal on a plate, you can attach both photos and Sayved reads them together as one entry. A single meal can hold up to two photos.
Yes. If the label lists protein, carbohydrate and fat, each is scaled by the same factor and stored with its own range. They roll up into the same totals as everything else you log, which you can see in the macro tracker view of your week. Sayved does not set macro targets, so nothing is compared against a plan. It adds up what you recorded.
Reading labels is part of Sayved Pro: 14 days free, then $2.99 a month. If you stop, what you have logged stays readable, searchable and exportable, free, forever. Your diary is stored on your phone with no account. The label photo is downscaled before it is read, nothing is kept on Sayved’s servers, and the original stays with your entry on the device. More is on the privacy page.
“Photo of a muesli bar label, then 'I had two of them'”
Illustrative entry. The range is narrow because the figures are printed, not judged by eye.
Yes. Sayved reads either, and scales the figures by the amount you give it. With no amount it assumes one label serving and notes that at low confidence.
Because the label's numbers are not the weak point. Without a stated amount, the guess is how much you ate, and a photo of packaging cannot answer that.
No. You just take the photo. Sayved works out whether it is looking at a plate, a label with an amount, or a label without one.
Sayved does not scan barcodes. It reads the printed nutrition table from a photo.
Yes, the original is stored with the entry on your phone. A downscaled copy is sent to an AI model for reading, with no name or account attached.
Photograph your plate and Sayved estimates calories and macros in seconds, always with a range and a confidence. Edit any number you disagree with.
Log protein, carbs and fat by photo, voice or nutrition label. Sayved estimates each with a range and shows your split. Every number is editable.
A voice food diary that turns "two eggs and toast" into a logged meal with calories and macros. Speech stays on your phone; audio is never uploaded.
A food diary app where you snap a plate or say what you ate. Sayved saves the meal with a calorie and macro range, and you can edit any number.
A supplement tracker app that logs what you took and how much from one spoken sentence or a bottle photo, then keeps a dose history you can ask about.
Free for 14 days, then $2.99 a month. Everything you've logged stays yours, readable and free, forever.
Sayved