Renowned Japanese Cake Store Biscuit Deploys Viscovery Visual AI Checkout for Digital Transformation

March 11, 2021

Biscuit (ビスキュイ), a popular cake store located in Shin-Shibamata, Tokyo, Japan, recently adopted Viscovery’s Visual AI Checkout system to enhance operational efficiency and improve customer services.

Exquisite Cakes in a Santa Fe Style Castle
Established in 1996, Biscuit has earned a reputation for its exquisite and delicious cakes, cookies, cake rolls, macarons and more. Taking design cues from Santa Fe, Biscuit provides a castle-like cake shop and a café for the enjoyment of cakes and afternoon tea.

Enhance Checkout Efficiency and Understand Customers Preferences with AI
As Biscuit’s business kept soaring, it encountered a problem and was eager to solve it. Since cakes sold in-store are freshly baked and do not come with any labels to have individual product sales data, cashiers would need to recognize each cake and enter the corresponding products in POS manually. Given the large volume of business each day and not wanting customers to wait for checkout, store clerks had no choice but simply just enter the price of cake products to speed up the process. As result of this, Biscuit did not have sales data of individual cakes needed for planning, marketing and managing store operations.

To ensure individual product sales data is recorded without slowing down checkouts, Biscuit decided to deploy ninascan, an AI-powered Visual Checkout system co-developed by Viscovery and Annie Corp. With this new Visual Checkout system, cashiers only need to press a button for the computer to recognize multiple products within 1 second, effectively shortening checkout time while allowing sales data of individual cake products to be recorded.

The sales data obtained from using AI-powered Visual Checkout system allows Biscuit to better obtain customer insights, observe customers’ preference changes for cakes, and develop more popular cakes.

Smartly Recognize Cakes in the Way Human Do
Using computer vision, AI and deep learning, the AI-powered Visual Checkout system is trained to recognize products like the way humans do regardless of products’ natural variance in appearance. Take cake as an example, the amount of topping, the size of blueberries, the position of chocolate flakes on a cake and the bottom holder under the cake vary, but the natural variance would not affect the recognition. Even with cakes tightly packed, the system is able to determine the correct type and quantity in a second. The Visual Checkout system helps Biscuit reduce approximately 45%~50% checkout time, eliminates the anxiety derived from long checkout queues, and improves shopping experience. Biscuit sales deputy director, Sugou Emi, indicates, “The system is even more efficient with larger number of checkout products.”

The Best Assistant for Rush Hours and New Cashiers
The Visual Checkout system has proven to be very beneficial for Biscuit. Currently, hundreds of transactions are processed through the system every day. Biscuit sales director, Yamaguchi Yuzu, shared, “When it comes to Christmas or Valentine’s Day, we experience double if not triple volume of customers and sales. With the AI-powered Visual Checkout system, cashiers are not only able to check out much faster but also with ease. We effectively hand over the repetitive recognition task to the computer allowing our store clerks to focus on in-store service enhancing customer satisfaction.”

Since the Visual Checkout system simplifies the checkout process with easy steps, both experienced and novice cashiers can be onboarded faster, reducing the time and effort to train a new cashier. The training time has been condensed from one month to two weeks. The system also further reduces errors from manual recognition.

With the development of AI technology, more and more retailers are seeking AI for operational optimization and digital transformation.

▲ AI-enabled Visual Checkout system recognizes products in various packages
▲ Even with cakes tightly packed, the system is able to determine the correct type and quantity in a second