MONDAY 5:25AM, I woke to my blaring alarm and dark Seattle skies at the University of Washington. I hopped on the Link light rail and zipped downtown before walking over to Amazon’s sprawling glass towered campus and arriving just outside my destination: Amazon Go, the much buzzed about vision of a tech-driven retail future.
The lights were on — the store’s distinctive signage glowing in the early morning. Inside the storefront, a small commercial kitchen was bustling with activity as chefs prepared food while workers inside the main store moved left and right.
Outside, I was met with a modest crowd of waiting customers and news crews. The mix of people was telling of how widespread the store’s intrigue was, with self-described Amazon enthusiasts, journalists, students, and locals who worked in other reaches of Seattle’s tech scene alike all waiting to see a first glimpse of our machine learning infused future.
When 7AM drew close, the doors slid open and out poured orange clad store ambassadors armed with similarly branded totes for our shopping use. We eagerly streamed in, scanning our QR codes from the Amazon Go app at the subway-like entrance gates to the store.
Directly in front of the store’s entrance stood rows of grab-and-go selections, presumably made in the front kitchen. Looking above, I was taken back by the sheer multitude of gray square cameras embedded across the ceiling and pointed to the various shelves, tracking people who picked up items or put them back. The sensor and camera ladened store no doubt provides a massive stream of data for Amazon Go’s backend computers to analyze, forming its aptly titled “Just Walk Out” technology.
Although Amazon originally described the tech as also using physical RFID tags for items in its patent filling, I noticed that the store makes no use of them, instead relying entirely on its machine learning capabilities. Surely, Amazon has been testing the limits of its tech, and with the volumes of people now filtering through the store, it remains to be seen how much more accurate and fault tolerant the system may become.
Of course, with the store’s abundance of sensors that track movements and learn people’s preferences through analysis, some view it with open privacy concerns. At the same time, it’s difficult for me to see a different way where the convenience of walking in and out of a store without stopping to checkout could be implemented.
Such privacy concerns are ultimately subjective, as Rebekah Denn writes in The Washington Post of her own “Orwellian angst” towards the store, while her “teenage son couldn’t care less.”
While viewing the food selection, I toured the rest of the Amazon Go store, tucked towards the right of the entrance. Snacks lined the walls along with Amazon branded meal kits, divided by a column with Amazon Go merchandise and a corner enclave for beer and wine that was guarded by a stationed employee checking photo ID.
Noticeably, Amazon Go does not carry produce or make room for a traditional deli. While not including such items certainly made implementing the Amazon Go concept store easier, ultimately, the limited inventory of goods and small space all reduce costs for Amazon to operate the store.
Additionally, Amazon’s Just Walk Out technology eliminates the conventional row of checkout lanes and cashiers at the front of grocery stores. I almost didn’t recognize their absence in the moment due to the store’s design that makes full use of the freed space.
Thus, much of the design for Amazon Go seemed to be characteristic of Amazon’s aim to remove friction from the buying process as it sets out to build a new model for grocery and retail stores.
The effect is saved costs for Amazon and time for customers — benefits that Amazon wagers people will become accustomed to as the store’s effortless shopping experience transitions from novelty to convenience.
However, while many stores currently have self-checkout lanes for buying items, the removal of checkout kiosks altogether from retail calls into question the future role of the 3.5 million cashiers working in the U.S.
At launch, I saw plenty of employees present at Amazon Go to welcome customers, oversee the store, stock shelves, and for those in the kitchen, prepare food items. Yet with the potential for technology similar to Amazon’s Just Walk Out system to be applied to established retail chains, including the company’s newly purchased Whole Foods Market, many theorize an upcoming widespread shift in retail jobs.
Having taken in the entirety of Amazon’s tech-driven retail experience, I grabbed a sandwich and salad to go, placing it in the blazing orange tote.
I moved through the groups of photo and video takers towards the exit gates, cognizant that my movements had been analyzed in near real time, enabling me to “just walk out.” Nonetheless, when I arrived by the exit, I paused for a second, with my mind telling me that there should be some kind of extra step in the whole process, before the gates swung open.
As I walked back to the Link station, past Amazon’s brightly lit glass Spheres, my phone buzzed with a notification from the Amazon Go app, announcing that the store’s sensors and data-analyzing algorithms had concluded what I bought that morning. I was presented with a receipt that listed all of the items I had placed in my bag and even gave me the ability to request an item’s refund in the off chance that I was charged incorrectly. Before pocketing my phone once more, I glanced at the top of the receipt for the duration of my trip: a cool 11 minutes and 18 seconds (I’ll work on that).
Overall, Amazon Go stands as a compelling example for how technology can and will be applied to our physical spaces. As the weeks and months go by, there will undoubtedly be plenty of discussion surrounding Amazon’s retail strategy, the impact of automation on jobs, and the growing role of AI in our lives.
But if there is one thing for certain from my morning trip to Amazon Go, it’s that the future is open for business.
Abhishek Joshi is a student developer at the University of Washington.