“There’s two problems in the retail industry. The experience for customers sucks. Secondly, it’s very hard to process all of the sales information in real time,” said Vasco Portugal, CEO and co-founder of Sensei.

I’d say the first problem largely revolves around having employees who are overworked and underpaid. Keeping enough staff around that checkout moves quickly and paying them enough that they don’t visibly hate their job can go a long way toward solving that. This is a customer service issue and removing employees is basically never a solution to improve customer service, except perhaps for specific bad employees.

For the second problem, I don’t see how the added data overhead of AI/Machine Learning helps improve processing time. This also doesn’t seem like a real problem to me. Processing transactions from a dozen or so registers simultaneously really shouldn’t be beyond the capabilities of a modern system, or even many old systems. The real issue I’ve seen is a lot of stores don’t seem to have good information on how much inventory is on their shelves, especially when stocking. The other challenge is the delay between someone putting something in their cart and reaching the register. I’m not sure how much of a problem that really is, though; stores can just display an alert if stock is low and if it’s gone from the shelf when someone gets there, too bad. You could put a barcode scanner on people’s carts that wirelessly updates your inventory and lets customers see their running total, but a lot of stores wouldn’t want that because it would probably reduce impulse purchases; it’s a lot easier psychologically for customers to put items back at the shelf than at the register.

This seems like a “solution” using the latest buzzword in search of a problem.