Why I stopped scanning barcodes (and why you should too)

For years, nutrition apps taught us that the best (and often only) way to log our diet was by hunting down barcodes in every corner of the kitchen or supermarket.

The process is well known: you grab a package, open the app, try to get the camera to focus on a tiny spot on the box, and cross your fingers that the information loads correctly. Although this was a revolution back in 2010, today it is a logistically obsolete method with serious adherence and health drawbacks.

That frustration was one of the main reasons we created Tabeku.

The bias toward ultra-processed foods

The biggest problem with the barcode scanner is psychological. Our brains always look for the path of least resistance.

A fresh apple, a pear, or salmon from the fishmonger do not have QR codes printed on their skin. However, any ultra-processed protein bar or industrialized sauce has one prominently displayed. If you log your food based on what is “easy to scan,” you will end up increasing your consumption of commercial products simply to avoid the laborious manual process of searching for raw foods in a database.

With the visual estimation based on Artificial Intelligence models that we use at Tabeku, this bias completely disappears. The app takes exactly the same amount of time to process a portion of homemade lentils or a banana through a photo as it would to process a box of pastries. Fast logging is finally democratized for real food.

The anarchy of crowdsourced databases

If the previous point wasn’t enough, there is a second colossal problem: user-generated databases.

We have all scanned a generic yogurt or a can of beans only to hit “accept” and realize a user entered the data backward or uploaded an entry of 6,000 kcal for a sugary soda. Conventional databases house decades of transcription errors with no professional validation.

The result? You waste even more time manually confirming and cross-referencing the physical nutritional labels with what your phone says. It’s a repetitive task that tires out anyone.

Delegating recognition to a highly trained, standardized AI focused on the visual aspect is the definitive leap that allows us to forget about bad math. It’s time to leave barcodes in the past.