The cashierless Amazon Go store is open for business, but it is still very much an experiment that has already faced numerous issues and delays.
But whatever the fate of the Go store, the influence of the Amazon brand has sparked a burgeoning market of technology developers lining up to offer retailers their own version of Amazon Go. Ideally, these offerings would learn from Amazon's mistakes so that other retailers can offer a smoother experience.
"We're building something that's more general and scalable, that can handle different cases and types of stores," said Steve Gu, the CEO of AiFi, which has attracted about $4 million in seed funding for its computer-vision technology.
The Santa Clara, Calif.-based AiFi is testing its cashierless technology with a retailer it did not identify. The vendor plans to run tests at other retailers and open a concept store ahead of a more formal rollout by the end of the year.
AiFi uses artificial intelligence, sensors and cameras to outfit stores with a checkout-free system that can deploy without regard to the differences in store sizes. It also uses algorithms and mobile devices to power behavior tracking and product recognition; this can be used for payments, inventory management and to gain insights into shopping habits and product preferences.
This technology can recognize groups of people who are shopping together and identify gestures, Gu claims. AiFi installs a network of cameras that communicate with each other to track shopping, and the retrieval of items from the shelf and in a shopping cart.
"The checkout is the same concept at Amazon; the sensors track the entire shopping session, identify who you are and charge your account at the store through your mobile app," Gu said.
The move to reduce or remove static cashier stations is picking up steam, said Ray Pucci, associate director of research and consulting services for Mercator Advisory Group, noting mobile self-checkout systems are operational in several retail stores, including Walmart and Sam’s Club.
"Meanwhile several fintechs are working on models for other merchants as well," Pucci said. "Most solutions utilize bar code scanning and payment via smartphone app. Amazon Go takes a great leap forward and eliminates the scanning of every item; that’s why the system has taken so long to perfect, and then only in a small footprint environment."
The use of machine learning informs the system, enabling it to improve its ability to recognize consumers, shopping gestures and volume of in-store items over time, which also helps it work in large stores with large crowds, Gu said. While Amazon has not commented much on the underlying technology behind Go, one of the challenges it reportedly faced was the sensors' ability to keep track of shoppers when there are large crowds in the store.
"The store we're testing with has tens of thousands of square feet," Gu said.
AiFi plans to charge a subscription fee, which Gu said is the best option to manage the cost of the hardware and updates to the store's software.
AiFi's not alone, as Amazon Go's potential has drawn other developers. Activant Capital recently was the lead on a $50 million investment in NewStore, a mobile retail platform. Another Activant portfolio company is RetailNext, which has roots in casino cameras and is extending its sensor technology for in-store use.
There's also Standard Cognition, which is also building no-cashier sensor technology to enable product and shopper recognition, along with payments.
"The simple reality is that retailers are competing with online sellers as much as they are with in-store environments," said Rick Oglesby, founder and president of AZ Payments Group. "So you must have a reason for consumers to come to your store in the first place, and you need to make it a good experience for them or they won’t come back the next time. At the same time, when consumers do leave their homes to shop, you can bet that they’ll prefer the store where they don’t have to wait in line to check out."