[AI Revolution](51) Three Missing Elements in Fully Unmanned Stores... Distribution Innovation with AI

FindersAI Opens Fully Unmanned Store 'Super Swift'
99% Accuracy in Product Recognition Without Scanning Barcodes
"Learning from Unmanned Stores Worldwide... Aiming to Be Korea's Amazon Go"

On the 26th of last month, the unmanned store 'Super Swift' opened in Seocho-gu, Seoul. I picked a few drinks and stood in front of the kiosk. Without even scanning the barcode reader, the list of purchased items and the payment amount appeared. It even did not miss the snacks I briefly put in my bag due to lack of hands. The accuracy of automatically recognizing the products selected by customers is 99%. This is possible only with cameras, sensors, and artificial intelligence (AI). Through this, AI startup FindersAI has automated everything from store operation to inventory management and payment.

Fully unmanned store Super Swift located in Seocho-gu, Seoul
Photo by Finders AI

Applying Vision AI Instead of Relying on 'Honest Payment'

There are three things that Super Swift does not have. Of course, there are no cashiers. There are also no expensive 3D cameras or LiDAR to replace the cashier's eyes. It uses 2D cameras and predicts three-dimensional movements not captured by these cameras with vision AI. Vision AI is AI that sees through cameras and judges like a human eye and brain. It also adds information detected by sensors that sense weight changes on the shelves. The AI combined with cameras improves accuracy and performs better than the same number of LiDARs. The cost of building such a solution in a convenience store of 30 to 60 square meters is less than 200 million KRW. Compared to global unmanned stores like Amazon Go in the U.S., it is about one-third the cost. Considering that labor costs alone for operating a 24-hour convenience store based on this year's minimum wage amount to about 100 million KRW annually, the cost efficiency is high.

Thirdly, there is no need for self-checkout payment methods such as kiosks or smart vending machines. Self-checkout requires consumers to scan the product barcode to proceed to payment, relying on 'honesty.' In other words, there is always a risk of theft. Even if payment errors occur, there is no way to respond. This is why owners of partially automated unmanned stores commonly stay nearby to manage the store. Ham Myung-won, CEO of FindersAI, explained, "Partial automation relying on CCTV cameras has low cost-saving effects," adding, "Most store owners operate several stores with a small margin and watch CCTV all day." In contrast, AI sends immediate alerts when exceptional situations occur. It also informs the inventory status in real time. This method reduces the owner's cost burden while maximizing consumer convenience.

Ham Myung-won, CEO of Finders AI [Photo by Finders AI]

To realize such a store, CEO Ham traveled around unmanned stores worldwide. He visited dozens of Amazon Go stores across the United States. While researching market reactions at stores in five European countries including the UK, France, and Poland, he was kicked out several times. By taking what was useful and discarding what was not from overseas stores, he completed the business model in three years. CEO Ham emphasized, "I even visited the same stores annually to learn technological changes," adding, "I concluded that we must create stores that are not only convenient but also cost-effective."

Customer Churn Rate Near 0%... Targeting Convenience Stores and Drugstores

Super Swift, operated for a month, received positive responses. About 100 to 200 groups visit daily, and only two groups left without entering. Major convenience stores such as CU, GS25, Seven Eleven, and Emart24 all visited. There are plans to open a collaborative store with a large retailer by the end of this year.

The goal is to expand the unmanned store sector starting with 60,000 convenience stores nationwide. They are targeting cafes, drugstores, and corporate cafeterias. They also plan to target Japanese convenience stores, which have a distribution environment similar to Korea. CEO Ham predicted, "With the labor population decreasing, it will be impossible to find convenience store part-timers in 10 years," adding, "Beyond labor cost savings, unmanned operation is the only solution."

In the mid to long term, they plan to focus on customer data analysis. They will analyze customer behavior such as how customers move around the store and which products they are interested in, using this data for product display and marketing strategies. CEO Ham said, "Until now, manufacturers have relied only on sales data," adding, "We will analyze the effect by understanding how much attention consumers pay when changing the design of drink caps."

This content was produced with the assistance of AI translation services.