Today, for lunch, I stopped in the new Amazon mini market in the north of downtown Seattle. I take a tuna sandwich in the fridges, I put it in my pocket and head for the exit showing a head as relaxed as possible. On the way, one of the employees stops me. I pause for a few seconds and the clerk says, “You should try the Mediterranean salad, it’s very good. ”
No, this store does not offer tuna sandwiches. This is Amazon’s brand new concept: Amazon Go, which was designed to avoid queuing at the cash desk. It is also an additional arrow to the arc of the sales giant and data service that he could uncheck to shake practices a little-outdated sector.
Small shop, big impact
To get into the store, I first downloaded the Amazon Go app on my smartphone. I connected it to my Amazon account and chose a payment method. The application generated a QR code on my phone screen that I scanned on a terminal in front of the store. Ceiling-mounted cameras and sensors followed all the items I put in my pockets. The app then automatically charged me these items when I walked out of the store.
The Amazon Go store has been in private beta for almost a year. Employees on the campus of the nearby company played guinea pigs. On Monday, January 22, it opened to the public for the first time, attracting a compact crowd in the premises of 160 m² located on the ground floor of the “Day 1” (the skyscraper of the company).
You can find everything you usually want in this kind of supermarket: takeaway food, milk, yogurt, coffee, snacks, and cereals. There is also a beer and wine department.
Unlike Amazon’s bookstores and electronics stores, where customers need to scan items for pricing and where to go to checkout to pay for purchases, Amazon Go lists prices for items in the shelves. Although there is no cash in the store, many employees are present to prepare the sandwiches or help the customers.
Everything is in the intelligence
But what happens with Amazon Go? Would the sales giant have decided to attack the mini-markets? It’s always a bad idea to underestimate Amazon’s ambitions. In fact, Amazon is hiding technological advances that will have major consequences in its sector of activity.
Amazon describes the technology used in its store as “computer vision, sensor fusion, and deep learning”. That is, in addition to tracking customers in the store, the system is able to associate you with what you buy. He must be able to understand that you are taking a yogurt, not the person standing next to you. He must understand that, finally, you will not take this packet of chips when you put it back on the shelves after putting it in your basket. Then he must be able to charge you and not the person who came out of the store a few seconds before you. If the system makes a mistake, Amazon would also be good at hiring a cashier.
This point of view is that of the customer. But, this system is obviously more complex because it must guess which customer buys such a product at this time of day, when replenishing stocks or shelves, how to present the goods to capture the customer’s attention, how to optimize the flow of pass through the store and perhaps even, how to stop troublemakers and shoplifters.
Of course, Amazon is not just a seller. It is a huge data warehouse that has built one of the largest cloud computing platforms (Amazon Web Services), which represents a growing portion of the company’s revenue. This platform is much more profitable than the low margins of the sales sector.
All the data and technologies that allow us to do our shopping without going to the checkout have many applications within the sales sector. These technologies must be trained in order to function properly. They need real data to build and optimize models because their role is very complex.
Even the AI needs training
I spoke with Daniela Braga, a renowned expert in data science and more specifically in AI systems training. She is also the founder of DefinedCrowd. Although she is not collaborating with Amazon on this project, she says that Amazon Go’s capabilities and experience are at the forefront of Learning and AI.
“The computer vision is not enough, adds Daniela Braga. Even though cameras, sensors, relays and other systems are coordinated, it’s a feat to be able to link all this data in real time. ”
However, one of the most important aspects of AI and machine learning is the ability of the system to learn and adapt. The more the system can observe its action models, the more it can refine its models to generate more efficient results and avoid errors.
Most organizations working on AI-based solutions do not have databases that are relevant to this application domain. Most vendors do not have access to the largest data clouds and processing capabilities.
But how far will it go?
Just like the Amazon library, Amazon Go seems to be a concept in itself that takes advantage of Amazon, the company that wants to make everything faster, cheaper and more convenient. And that allows us to question ourselves about the future.
Will the company expand its cashless store concept to its other brands (including Whole Foods)?
Is Amazon simply the first customer of its own technology that it will then offer to its competitors? The company has launched Alexa on its Echo devices but is now selling it to other manufacturers, including some rivals. It is indeed more profitable to own them than to monopolize the means of distributing them. This also applies to large retailers.
Is this an aspect of the city of the future, for which Amazon has been working for a long time? In Seattle, Amazon executives have in front of the new generation of urban millennials who have finished repaying their student loans and who are housed in apartments barely larger than shoe boxes, which do not have a lot of storage and even less space for products purchased wholesale at Costco. Top-of-the-range mini-stores like Amazon Go correspond more to this population than traditional supermarkets. This new concept is, therefore, an opportunity worth taking.
Will there be more customization in the future? The next time I go to Amazon Go, I may be greeted by a hologram and will ask me this time, I’ll take a salad Nice? But that’s not just about Amazon.
Amazon has not revealed its future plans, so we will have to draw our own conclusions. But the real question is: how much are we willing to pay to buy a tuna sandwich without waiting in line at the checkout?