You're in the aisle with a box in your hand, and an app has just given it a number. The number doesn't know you have Hashimoto's. It doesn't know what you scanned last week, or that you've now read this same label three times because something in it looked wrong to you and you can't say why. It's the same number it would hand anyone.
Short answer: Yes, a food scanner can be personal to you. What makes it personal is context: the condition you're managing, what you've scanned before, and what else you're already recording. A personal scanner names the ingredient it flags and the reason it flags it, in plain words, and it shows you something better in the same aisle. That last part matters more than people expect. A flag with no alternative is a dead end.
What should a personalized food scanner do?
Six things. Miss one and you're back to a number.
- Read the back of the pack. The claims on the front are marketing. The list on the back is the product.
- Name the ingredient and the reason. "Contains sunflower oil, an industrially extracted seed oil we flag" tells you something. A number out of a hundred doesn't.
- Show the gradient. Sugar is sugar, but a whole date isn't a spoon of corn syrup. A scanner that scores granola with a little cane sugar the same as a soda has just told you it isn't reading carefully, and you'll stop believing it.
- Know which condition it's reading for. A scanner built for general clean eating and a scanner built for a thyroid condition are looking for different things on the same label.
- Offer a swap you can actually buy. In the same aisle, in the same category, today. A recipe or a general principle won't help you at the shelf.
- Say when it doesn't know. Thin coverage stated out loud beats a confident answer built on nothing.
That's the whole test.
Why does one score for everyone fall down?
Because a single grade is a population instrument, and you're one person.
That's not a marketing position, it's how the people who build these scoring systems describe them. A 2026 review of Nutri-Score and related front-of-pack labeling systems in the journal Nutrients puts it plainly: "Current nutrient profiling systems are designed for population-level recommendations and implicitly assume that foods exert broadly similar effects across individuals." The same review says these indices "should be interpreted as population-level indicators of dietary quality rather than as direct predictors of individual disease risk". A tool designed to describe a country is being handed to you in a supermarket.
The apps built on top of that don't agree with each other either. Lindsey Wohlford, a Wellness Dietitian at MD Anderson Cancer Center, writing in November 2025, notes that scanner apps "use a variety of methods to assess how healthy different foods are" and that "these methods can vary widely from app to app". Her sharper point is the one you've probably felt already: "The primary downside of these apps is that they oversimplify nutrition. An app generally rates foods as 'good' or 'bad' based on its own criteria." She calls it "a very black and white view of food that does not account for such nuances in diet."
You already know this. It's the moment the granola and the soda come back looking similar, and you quietly stop trusting the app. That reaction is correct. The tool flattened something that wasn't flat.
If you want the longer version of that comparison, we wrote it up in is there a Yuka alternative that reads labels for your thyroid.
What does "personal to you" actually mean?
It means the app knows who it's reading the label for.
Osmi reads the product against you: your symptoms, your scan history, your cycle, your sleep. Instead of a generic score, you get a plain answer about those ingredients through the lens of your thyroid condition, with the reasoning spelled out. The ingredients didn't move. Your context did.
Two things to be honest about here, because the word "personalized" gets stretched.
First, this is context rather than prediction. Osmi isn't telling you how a food will affect you, and it never connects something you ate to how you felt or to a number on your last panel. It doesn't diagnose, it doesn't grade you, and it will never ask you to change your medication. What it does is read a label with your condition already in the room, which is a different job from scoring a product for everybody.
Second, the personal part gets better with use, because Osmi keeps your scan history and reads each new product against it, and against the rest of your record. Your labs move two to four times a year. How you feel moves daily. What you buy moves weekly. A scan history is only worth something read next to the rest of what you're keeping, which is what your thyroid record is for.
What does a useful flag look like?
Here's the shape of it, in the app's own words.
You scan a carton of barista oat milk. Osmi reads the ingredient list for you and gives you three rows:
- Added sugar. Avoid.
- Rapeseed oil. Avoid.
- May contain traces of gluten. Watch.
Then, underneath, a cleaner swap: oat milk, oats and water only. Better.
Each flag carries its reason rather than a verdict on you. Written out, one of them reads: contains rapeseed oil, an industrially extracted seed oil we flag, often pressed with heat and solvents and oxidized by the time it reaches you. Swap: the same product made with extra-virgin olive oil, or with butter, ghee, tallow, coconut oil or avocado oil in the ingredients instead.
We'll say the uncomfortable part out loud, because a scanner that hides its position is only being vague about it. Mainstream guidance still calls seed oils heart-healthy. We disagree, on processing and oxidation grounds, and we tell you that's our position rather than dressing it as settled. Same with wheat, barley, rye and malt, which we flag for a Hashimoto's user when most guidance would only flag them with a celiac diagnosis. You're allowed to disagree with us. You can't disagree with a number that never explained itself.
The full ingredient lists live in the sibling pieces: an autoimmune food scanner for what we flag with a thyroid condition in mind, and a seed oil scanner for that one ingredient family.
Why does every flag need a cleaner swap?
Because a flag on its own leaves you standing there.
You've got twenty minutes and a product you suddenly don't want. Telling you it's bad has moved you backwards. You either buy it anyway and feel worse about it, or you put it back and go home without oat milk. Neither of those is help. So the rule is simple: every flag comes with something better, right next to what you scanned, in the same aisle. A flag with no alternative is a dead end.
This is also the honest answer to why scanning feels heavy at first. The flags were never the heavy part. The dead ends were. Once every flag arrives with a swap attached, the scan stops being a judgment and starts being a shortcut.
Why does saving the ingredient beat saving the product?
Because you're going to meet that ingredient again, in a different box.
Most apps let you save a product. That helps exactly once, at the shelf where you found it. Osmi saves the ingredient, so the next time it turns up, under a different brand or a different name, the answer is already yours. It's the difference between a shopping list and actually reading labels faster.
That's the quiet compounding part. Month one, you're scanning. Month six, you're recognizing.
What can no scanner tell you?
This is where we'd rather be blunt than impressive.
A label doesn't have to list everything. Under US labeling rules, 21 CFR 101.22 allows "spice, natural flavor, and artificial flavor" to "be declared as 'spice', 'natural flavor', or 'artificial flavor', or any combination thereof". A whole set of ingredients can legally arrive as two words. Any scanner reading the label inherits that ceiling, ours included.
A missing warning isn't a clean bill. The FDA requires the nine major food allergens to be declared: milk, eggs, fish, Crustacean shellfish, tree nuts, peanuts, wheat, soybeans and sesame. But the "may contain" and "produced in a facility" lines are different. The FDA says those advisory statements "are not required by law". So their presence tells you a manufacturer chose to warn you, and their absence tells you nothing at all.
Coverage is never complete. No scanner has good data on every product, and ours is no exception. When Osmi's read on a product is thin, it says so instead of guessing.
And a scanner is not care. Wohlford recommends avoiding these apps if you "need personalized nutrition guidance, such as if you're on a diet for certain medical conditions", if you "tend to feel guilt or anxiety about the foods you eat", or if you're "at a high risk of an eating disorder". We think that's right, and it's worth reading twice. She also warns that sorting food into good and bad "can contribute to anxiety, shame, guilt and rigid thinking about food". That's precisely why Osmi flags the product and never grades you. There's no daily food score, no streak, no verdict on how you did.
Her own advice for reading a package without any app is worth stealing, and it happens to be the principle underneath every flag we write: "When a product has more ingredients, it often means more processing", so "the fewer ingredients, the better".
That's the whole idea. The closer to the thing as it grows, the better, and everything else is detail.
Questions women ask about personalized food scanners
I want a food scanner that is personalised to me instead of giving everyone the same score. Does that exist? Yes. Osmi reads the product against you: your symptoms, your scan history, your cycle, your sleep. Instead of a generic score, you get a plain answer about those ingredients through the lens of your thyroid condition, with the reasoning spelled out. To be clear about what that is and isn't: it's context rather than prediction. Osmi doesn't tell you how a food will affect you and never links something you ate to how you felt or to a lab result. It reads the label with your condition already in the room.
Is there an app that suggests a cleaner swap when I scan a product? Yes. Every flag in Osmi comes with an alternative, right next to what you scanned, in the same aisle. Scan a barista oat milk and you'll see the added sugar, the rapeseed oil, and the traces-of-gluten line, then a cleaner swap: oat milk made of oats and water only. A flag with no alternative is a dead end, so you won't get one. Osmi also saves the ingredient rather than just the product, so the next time it turns up under a different brand you already have your answer.
What is the best food ingredient scanner app? Judge it on six things rather than on a ranking. Does it read the ingredient list, or just the front of the pack? Does it name the ingredient and the reason instead of handing you a number? Does it show the gradient, so a whole date isn't treated like a spoon of corn syrup? Does it know which condition it's reading for? Does it offer a swap you can buy in the same aisle? And does it say when its coverage is thin? An app that misses any of those is giving you a score, not an answer.
Which food scanning app is the most accurate? Accuracy is the wrong first question, because these apps aren't measuring the same thing. Lindsey Wohlford, a Wellness Dietitian at MD Anderson Cancer Center, writes that scanner apps "use a variety of methods to assess how healthy different foods are" and that "these methods can vary widely from app to app". Every one is also bounded by the label. US rules let "spice, natural flavor, and artificial flavor" stand in for a whole set of ingredients, so no scanner sees past that. Ask what an app reads, what it flags, and whether it says when it doesn't know.
Can I scan skincare and makeup as well as food? Yes, and they're kept separate, because what you eat and what goes on your skin reach you by different routes. Food and drinks are read for things like added sugars, seed oils and hidden gluten. Skincare and makeup are read for fragrance and the other ingredients we flag. Both come back the same way: the ingredient, the reason, and something better in the same aisle.
What happens if a product isn't recognized? Osmi tells you it doesn't have good data on that one, instead of guessing. Coverage is never complete for any scanner, and ours is no exception. A confident answer built on nothing is worse than a blank, so when our read on a product is thin we say so. You can still read the list yourself, and the rule that holds up without any app is the plain one: the fewer ingredients, the better.
Educational, not medical advice.