Dive Brief:

• Propulse, a company founded and led by former Saks Fifth Avenue executive Eric Brassard, has launched a product recommendation platform that relies on artificial intelligence, machine learning and image analysis to intuit what items should be recommended to shoppers to buy next.

• Propulse uses what it calls a Purchase Path Neural Network, employing proprietary deep learning and image recognition algorithms. The platform analyzes thousands of product image variables to determine what the shopper currently is looking for, and what other products they might want to purchase later.

• The platform also can help retailers understand which inventory they should stock more or less of by tracking the inventory status of products recommended to shoppers. If the product recommended turns out not to be in stock, Propulse logs the interaction and sends the data to the retailer.

Dive Insight:

AI and related technologies increasingly are making their way into retail at many different levels. In this case these technologies are being used behind the scenes to improve on product recommendation capabilities that were unheard of not so long ago, but have become stale in their own particular way.

Those recommendation engines are supposed to be helpful and just presumptive enough to help customers understand what the might need or want given what they just purchased. As Propulse points out, this is often presented quite literally with statements like “Because you bought this...” or “Customers who bought that also bought this...” recommendations.

Those approaches are what you might call superficial recommendations, and don't account for personal taste, whether or not shoppers already may have purchased the products they are recommended, or product images shoppers are looking at in real time. It seems like Propulse has designed a system that is meant to be more intuitive to the full range of shopper behavior, and not just reacting to the purchase itself.

Fashion retailer Frank + Oak has put Propulse to work and is seeing results, according to CEO and co-founder Ethan Song. ”With Propulse, we’re able to complement our democratized current personal stylist offer by showcasing new items to our customers that pair well with existing ones they selected — that’s what keeps them coming back for more and deepens our relationship with our community."

While AI can be used in retail in different ways, solutions like Propulse's and Cloverleaf's Emotion AI and whatever Blackbird Technologies is now doing for Etsy suggest a future in which retailers will be able to read customers minds — or facial expressions, or behavioral patterns — to influence the shopping journey in ways far more subtle than screaming marketing messages at them.

Propulse's take on this is especially intriguing because it is built on the experiences of a former retailer executive, one who sees among all of the advantages e-commerce has over brick-and-mortar retail at least one thing it doesn't have: In-store associates who can understand what shoppers are looking for, and give them personalized assistance.