Every day, millions of receipts are printed. Health, food, finance, these pieces of paper which may seem worthless, contain a vast amount of data. Behind these sometimes indecipherable product lines there is actually hidden information that is very profitable.
We are a start-up located in the Venice of the Alps 🏔, and our ambition is to make the most of the data with the latest cutting-edge technologies. The will to permanently challenge ourselves and our technologies together with our open-mindedness, it's in our veins. This is the force that drives us to improve on a daily basis and not remain complacent.
In the field of receipt analysis, current crowdsourcing solutions partially transcribe the receipt with an average of ~10 minutes per receipt. Understanding the content of a large amount of receipts by human operators is simply unrealistic, expensive and inaccurate. For the past 5 years, we have been working on solving this problem and we have found that the rate of errors committed by human operators when understanding receipts is around 20%. Today, the progress made in the world of artificial intelligence has allowed us to find a solution to all these problems.
If you thought all receipts share the same format you will be disappointed. It's even more striking if you take time to look at the products' labels.
CSHGRT DRNK BB CCNT seems incomprehensible to you?
We managed to understand this at scale.
It's actually a yoghurt drink with cashews, blueberries and coconuts! (by the way, if you guessed it, congratulations!)