New online shopping startup tries to predict what you'll want to buy next

Most online shopping sites are great at suggesting items similar to the ones you've previously purchased. But until now, none have quite mastered how to use that data to actually predict what you will be interested in next. That's the premise of Propulse, a new Montreal-based startup that uses machine learning and image recognition software to attempt to understand a shopper's tastes and recommend new products accordingly. SEE ALSO: Amazon's getting serious about its delivery ambitions with a $1.4 billion hub The e-commerce platform, which officially launched Thursday, is billed as a sort of online retail analog to Netflix's video recommendation engine. Its algorithm analyzes nearly 40,000 attributes of product images or listings in order to suss out a customer's personal style and match it to known shopping patterns.