Because food isn’t simple — and the way we talk about it should catch up.

Most food scanning tools are built on a tidy idea: scan a barcode, get a score, make a decision. That sounds helpful, and sometimes it is. It is a good rule of thumb to avoid items that have a bad score. However, saying that a product is “good” or “bad,” or giving it a numeric score, is rarely the full story. Spinach is healthy — until it’s all you eat. On top of this, some ultra-processed products sneak through with a “green” score because they meet narrow nutrient rules while ignoring ingredient quality or processing level. That’s not helpful. It’s misleading.

Ultra-processed foods are engineered to optimise for shelf life, taste, texture, cost and other factors. They can similarly be engineered to score well on any known scoring system. We already see evidence of this in “clean label” efforts, where products are ultra-processed to have labels that look simple and healthy (New Ingredients, New Processes: Managing Established Risks, 2024) (Kerry Group, 2025).

Instead of looking at the health scores of individual products, we should be looking at our whole diet over time. To observe trends, and to learn what to change next. Tracking our diets should be more than the micro or macro-nutrients. On top of all this, we should also know our percentage of utra-processed foods, we should be able to know what we eat: how many emulsifiers, colours, flavourings, gums, firming agents, sugars, artificial sweeteners, and so on we eat and where they come from.

Products can have multiple variants in the market at the same time. The name, branding, graphics, and barcode might be identical — but the ingredients and nutrition can vary dramatically. Existing apps typically track one variant for each product. We aim to track all variant that we find and keep a history of them. This will allow us to analyse trends later and provide more accurate product information right now. Read more about product variants.

The NOVA food groups weren’t designed as an individual product score. If you’re already trying to reduce ultra-processing you might’ve asked yourself if something is really UPF or not. For some items it was trivial. For others it was borderline and hard to tell. There is no single way to determine if a product is ultra-processed or not. Current apps and studies either look at the category of the food (eg protein snack bar -> ultra-processed) or go a step further and look at the ingredients. There is no clear answer or algorithm as the processing steps foods go through aren’t declared on labels. What we can do is iterate through new algorithms for assigning NOVA groups to food items and have a discussion pages for all products, where humans can discuss product tagging, alternatives and bring arguments for putting the product in different NOVA groups.

Lastly, where do you find better alternatives? Where do you actually buy them? We’re aiming to build a social feature where you can share non-ultra-processed you find and where you find them to help others eat better too.

Why yet another food scanning app?

Does any of this sound interesting?

References