Atos Helps French Startup Selartag Rapidly Build Wine Inventory Tool Using AWS

Taking an accurate inventory of a cellar at a high-end restaurant is an essential and regular but time-consuming task. It typically occupies two highly trained sommeliers, using pen and paper, for at least 2 days a month. In addition, the audit process can be prone to mistakes, with an estimated 10 percent margin of error in stocktaking.

Selartag wanted to develop a subscription service aimed at restaurants and private wine collectors, vineyards, wine dealers, and distributors. The company’s founders wanted to give its customers the ability to track individual bottles of wine, but they also realized that technology could offer a great deal more. The goal was to create a system that can make the whole wine supply chain ready for a variety of services already common in other types of logistics and distribution, such as just-in-time inventory control and intelligent stock control.

These give their customers data-rich insights into consumption and offer advisory services for restaurant staff and individuals, such as providing details on the wine’s provenance, food pairing suggestions, and other advice.

Selartag began working with Atos in November 2022. “I knew Atos’s reputation through their work with my previous employer,” says Herve Lemaire, president and founder of Selartag. “They might be seen as too big for a small startup like us, but I trusted them, and they believed in my product.” Lemaire appreciated the flexibility and agility that AWS offered and knew that it could provide a tailored solution. “We know that, thanks to AWS and Atos, our application is ready to scale instantly when we sign up a large chain of restaurants or even a vineyard,” he says.

Selartag decided to use multi-function labels that use RFID so that they can be identified at a distance using a UHF (ultra-high frequency) scanner or QR codes. This would enable users to identify bottles using a mobile phone application. Sommeliers, or other staff, use a handheld scanner to carry out inventories of the wine. The scanner can simply be passed in front of the wine racks to detect the bottles. The scanner uses Wi-Fi to link to AWS and update the database. Selartag also offers a UHF gateway scanner so that busy restaurants can keep track of bottles leaving the cellar with no manual checks by waitstaff—the bottle is automatically logged as it passes through the gateway when it leaves the wine cellar.

Selartag’s technology can update in near real time to improve inventory speed and accuracy and can improve security by tracking individual bottles. By improving inventory accuracy, the restaurant has less chance of running out of popular wines or wasting money on stock that isn’t selling.

Some restaurant cellars contain up to 400,000 bottles, so individual scanning of bottles is not practical. The data is processed by Amazon Elastic Compute Cloud (Amazon EC2) instances for easily scalable compute power, and uses AWS Lambda, a serverless compute service, for the main architecture and the auto-mediation service procedure. The team uses Amazon Relational Database Service (Amazon RDS) for its Postgres database to store information about the bottles. With accurate data on stock levels, Selartag can use AI and Amazon Bedrock to provide a portfolio of additional services, such as more intelligent inventory advisory functions, or offering suggestions about which wines are likely to sell in the coming weeks based on historical data, weather, and other data sources.

The system is already working well in many restaurants, including Michelin-starred establishments. One of the prestigious chefs using Selartag described the solution as a “fantastic tool.” Using it reduces pressure on his staff and frees them up for more valuable tasks.

Using Selartag technology, the audit process can be accomplished by an untrained staff member in less than 30 minutes—up to 99 percent faster than the manual approach. The 10 percent margin of error in human audits has been reduced to less than one percent using the technology. This means that there’s less chance that expensive wines are left in storage too long and not drunk at their best.

The next capability to be added to the system is a find-a-bottle function, which uses the scanner to guide staff to a requested wine, wherever it might be in the cellar. Using Amazon Bedrock, Selartag will be able to match wines from an individual’s cellar with a planned menu and suggest new wines based on the user’s existing favorites. “We’re just getting started with the possibilities of this technology,” says Lemaire. “AI offers us so many potential possibilities for wine collectors, for restaurants, for vineyards, and for the whole wine supply chain.”