AI Revolutionizes Inventory: Vinasoy Slashes Out-of-Stock by 20% with AWS

Vinasoy, a prominent Vietnamese soy milk manufacturer, has revolutionized its retail display monitoring process by deploying an AWS-based generative AI system, developed in collaboration with Renova Cloud. This innovative solution has led to a remarkable 20 percent reduction in its out-of-stock rate within just two months of deployment, significantly enhancing operational efficiency and sales responsiveness across its vast retail network.

Prior to the implementation of this AI system, Vinasoy's sales teams encountered substantial delays in acquiring critical display compliance data. The manual assessment of store product-layout photographs, processed through the company's Distribution Management System (DMS), was a labor-intensive and time-consuming endeavor. Each scoring cycle demanded nearly 2,000 work hours and extended over 20 days to complete, covering only about 17 percent of outlets monthly. This slow, infrequent monitoring meant that by the time sales teams received the data, shelf problems had often already resulted in lost sales, as highlighted by Le Ba Be, National Sales Director at Vinasoy.

The new AWS-powered system streamlines the entire workflow. Salespeople photograph product displays during store visits and upload these images to the DMS. Amazon SageMaker then processes these images using Vinasoy's own advanced image recognition technology. This system is capable of identifying 24 distinct versions of Vinasoy milk, including various packaging and sizes, at an impressive rate of 22 images per second, which is approximately 1,300 times faster than manual image inspection. Following this, Amazon Bedrock takes over, powering the generative AI stage that evaluates each store's display against Vinasoy’s approved product display standards. The system demonstrates robust capabilities, assessing even imperfect photographs and unusual shelf arrangements while maintaining accuracy, though no specific numerical accuracy measure was provided in the announcement.

The deployment has yielded transformative results, dramatically improving Vinasoy's ability to manage product placement across its retail network spanning 34 provinces. Sales teams now receive accurate compliance scores within one to two days, representing up to a 20-fold increase in speed compared to the previous manual process. This expedited reporting allows sales staff to identify and address stock issues and non-compliant displays proactively, preventing potential lost sales. The monthly monitoring coverage has expanded significantly from 17 percent to over 70 percent of outlets, with a weekly monitoring frequency, enabling Vinasoy to maintain consistent product availability. Le Ba Be further emphasized the positive impact, stating, “This has freed our people to do what they do best: build relationships with retailers and keep our products within reach of consumers.”

Retail display monitoring is just one facet of Vinasoy’s broader strategic integration of AI. The company is also leveraging AI-driven data analytics to optimize sales management, including demand forecasting and distribution. As an example of improved efficiency, preparing a sales report for management meetings now takes approximately five minutes, a stark contrast to the previous three days. Vinasoy plans to expand its display monitoring system to cover all its outlets in the coming months and is actively collaborating with Renova Cloud to develop an AI-based packaging inspection solution on AWS. Eric Yeo, Country General Manager for AWS Vietnam, lauded Vinasoy’s initiatives, noting that their journey exemplifies how FMCG companies can transition from AI experimentation to achieving meaningful business impact.