The factories that process our food and beverages (newsflash: no, it doesn’t come straight from a farm) have to be kept very clean, or we’d all get very ill, to be blunt. Ensuring that usually entails deploying petri-dish-based microbiological monitoring, hardware and waiting for tests to return from labs. A new startup has plans to use deep-learning algorithms to speed up this process.
Spore.Bio is a French startup that has developed a new pathogen-detection methodology. It works by shining an optical light on surfaces where clean food has been, and doing the same with unclean food. It then compares the two datasets to detect when a surface is not clean.
Off the back of this solution, it’s now raised €8 million in pre-seed funding led by London’s LocalGlobe VC.
The images that Spore.Bio produces are being read beyond what the naked eye can see. “We have machine learning models that will recognize the spectral nature of the bacteria in this snapshot. To make our system work, we have to train it with lots of samples of foods and beverages, contaminated and non-contaminated, to create this huge dataset. That is a huge asset for us. That’s why we signed some contracts with some of the biggest manufacturers in the world.”
Spore.Bio is a startup very much still in its early stages. The raise will be used in part to work on a device to handle this monitoring more easily.
“We’re building a hardware device that is able to detect pathogens immediately, directly on the factory floor. This handheld device makes it easier to carry out quality sampling, providing almost real-time insights into any potential bacteria in the factory,” Amine Raji, the CEO claimed.
Spore.Bio claims its solution will eventually work almost in real time. The implications are that a food processor will end up with less down-time. And that is significant, because according to research by Deloitte, the cost of downtime to the global food and beverage processing industry alone is estimated to be in the region of $50 billion annually. (Of course, Deloitte has some skin in the game, so take its big number with a grain of salt.)
Although the the product isn’t currently live, Raji said it has a “waitlist” for its first prototypes, which they are hoping to deploy globally by next year to clients’ sites.