Eat with your
eyes open.

Svelio is a nutrition app that reads the actual food label, grades every nutrient against FDA Daily Values, flags additives, checks marketing claims against the ingredient list, and grounds an AI nutrition coach in your own logged data. It runs on the web, iOS, and Android.

Photo-calorie apps guess your calories from the pixels in one picture, and you have no way to check the number. Svelio works differently at every step. When there's a label or barcode, it reads the manufacturer's own published numbers instead of guessing, then grades each nutrient against FDA Daily Values with fixed rules, so the same product always scores the same way. When all you have is a plate of food, Svelio still anchors every item to USDA reference data rather than eyeballing calories, runs a second pass that re-checks its own portion estimates, and lets you shoot up to three angles of the same meal for a sharper read. And because it reads the ingredient list too, it flags additives and catches marketing-versus-reality gaps that an image-only calorie app can never see. Full breakdown at svelio.io/competition.

Svelio flags more than 40 additives with plain-English explanations and the history of where they have been banned or restricted. Risk levels come from a deterministic ingredient database, not AI guesses, so the same ingredient always gets the same assessment.

The coach answers questions about your own logged food and patterns. It is grounded in a live snapshot of your data and protected by a multi-layer hallucination defense, so it never fabricates what you ate or claims actions it did not take.

Yes. Svelio cross-checks front-of-pack claims against the actual ingredients and nutrition, flagging gaps like "All Natural" with an artificial color, "No Added Sugar" with high-fructose corn syrup, or "Healthy" with very high sodium.

Svelio is tested against ground truth rather than against itself. Photo estimates are checked by weighing the food on a kitchen scale, and where the food carries a printed nutrition label, the label is the reference instead of someone's guess. That testing drives a deliberate design choice. Research from the NIH's National Institute of Diabetes and Digestive and Kidney Diseases, using meals prepared in the NIH Clinical Center's metabolic kitchen and presented at NUTRITION 2026, found that photo-based calorie apps systematically under-count: they missed roughly 250 to 345 calories and about 30 grams of fat per meal on average, and did worst on higher-fat food (sciencedaily.com/releases/2026/07/260726015237.htm). Svelio's design targets that specific failure: calorie-density floors stop a rich dish being scored lighter than it physically can be, and cooking oil or butter that a single photo hides is surfaced as its own line rather than dropped.

That is a design choice, not a guarantee. Some foods are genuinely hard to size from a photo, smooth or blended ones like a plain sauce most of all, and no app reads a portion perfectly from an image. So when a portion cannot be read reliably Svelio marks it as an estimate and leaves every item editable, instead of presenting a confident number it cannot stand behind.

Svelio's free tier includes unlimited label scans and five coach messages per day. Pro is $5.99 per month or $27.99 per year with a 7-day free trial, and there is no weekly-pricing trap.