Last week, from the 26th until the 29th of September, we joined another industry meetup: Alimentaria Foodtech in Barcelona. A lot of familiar faces, many news ones too, and some new ideas after talking to all of them.

One takeaway is the growing number of companies concerned about their powder food ingredients, whether they buy functional blends, spices blends, bakery blends and other, manufacture these blends, or even distribute powder products. They are all looking for better tools for quality control.

Once again, fraud is a growing concern, and it makes sense.When you deal with mostly powder products, with hard to describe differences, and easy to conceal adulterants in, it is a tempting product to mess with. Remember that an estimated $ 40 billion of fraudulent products move through our food system every year.

But on supplier monitoring, companies are also looking for better tools. The old tale of, “I don’t have to check you because I trust you”is no longer enough in a world where price and availability changes makes it essential to manage several suppliers to ensure business continuity. Better tools to check their quality are needed.

Even distributors are concerned about the ingredients they sell. Although they don’t manufacture anything, the chain of responsibility puts them in the crosshairs of potential litigation, that are a waste of resources and time, even when it turns out happily. Distributors need better ways to check that the raw materials are what the label indicates and not something else.

What can Chemometric Brain do for better food quality control?

Our fingerprint based qualitative analysis allows food companies to rapidly evaluate powder-based food ingredients, pure or blends.

Leveraging near-infrared spectroscopy – which generates a unique fingerprint of the product based on its chemical composition – we train models to recognize the particular ingredient.

The data used to train these models are samples previously determined as “good”, which will give the model an accurate representation of the product.

Once the model is trained and robust, unseen samples can be evaluated against it, telling you if it is what you expect, or something is different about it.

A close parallel is facial recognition, which you train using some scans of your face, and will only unlock your phone when it is you in front of it.

In short, it is a fast and sensitive technology to detect any minimal deviation in the composition of raw materials, blended ingredients and final food products.

‍