Artificial intelligence has, not unsurprisingly, been a headline topic at the NRF Big Show, and NVIDIA is undoubtedly one of the companies leading the industry’s innovation in this sector.
Typically known for its Graphics Processing Units (GPUs), which power breakthrough technologies for companies like Meta and Open AI, NVIDIA is now staking a claim in the retail industry by powering solutions for companies ranging from L’Oréal to Lowe’s and Walmart.
In many ways, this show was a big “coming out” for the company, which also unveiled the NVIDIA AI Blueprint for retail shopping assistants, a generative AI reference workflow designed to help developers create AI-powered digital assistants that work with and support human workers online and in stores.
Generative AI has “completely reinvented” ecommerce, according to Azita Martin, VP and General Manager of Retail & CPG for NVIDIA. Search has been especially impacted, with generative AI models, catalog enrichment, reasoning and embedded text supporting a much richer and more personalized experience where shoppers can find exactly what they’re looking for.
AI Use Cases Span Marketing, Service and Merchandising
However, AI also has the potential to influence different retail functions and teams, from marketing to store design and supply chain. In the case of L’Oréal, the company created digital 3D twins that it delivered to ad agencies, like its primary partner WPP, to use through other platforms like Stable Diffusion, so “creative teams could create ads and marketing campaigns much faster,” said Martin. The ability to generate creative faster and at scale enabled the brand to improve its marketing reach and impact across different channels, from social media and beyond.
Martin also offered the example of Lowe’s to showcase how retailers can harness the power of digital twins to support agility, efficiency and speed to market. The home improvement and DIY retailer created digital twins of 1,700 stores and is updating them several times a day with new operational and inventory data. “As a result, they’ve been able to simulate different layouts to really optimize how customers are shopping in stores, and how to change layouts to ultimately improve revenue,” Martin said. The retailer also is creating digital twins with their planograms to optimize their merchandising.
Finally, Walmart tapped NVIDIA to improve forecasting by injecting a large amount of data and sorting “hundreds of millions of combinations of SKUs and stores on a weekly basis,” Martin explained. By forecasting and running those algorithms more frequently, a company of Walmart’s size and scale can see significant results with merely a 1% improvement in forecasting.
Advertisement
Walmart also is in the process of implementing a shopping assistant, which essentially “takes your best and most knowledgeable associates and replicating that 24 x 7, at scale, and on the ecommerce site and mobile site,” Martin said. These assistants are not only supporting customers; they’re also empowering associates who need fast and easy access to information to better support store operations and service.
‘Physics AI’ and the Opportunity for Global Supply Chains
John Furner, President and CEO of Walmart U.S., who led the discussion with Martin, noted that inventory is the next frontier for AI. “Inventory is the asset that can make a customer experience fantastic, or it can hurt it due to mismanagement or a static supply chain that cannot dynamically respond or move inventory,” Furner said. “This technology is going to change the way we all manage our inventory.”
Martin agreed, indicating that supply chains will benefit the most from AI — specifically “Physics AI.” “That’s basically AI that understands physics, including the weight of things and the volume of items. Basically, it allows retailers to create a physically accurate digital twin of stores or distribution centers. We can simulate different layouts to analyze how people and objects behave within those different layouts.”
Making changes in layouts can be time-intensive and disruptive. Being able to simulate different layouts and operations before a change is made at the store or DC level can eliminate that disruption and associated costs, Martin explained. Ultimately, this can help retailers improve throughput and fulfill orders faster. These capabilities, paired with more granular forecasting, can ensure products get to the right customer at the right time, from the right source, whether it be a store, a distribution center or warehouse. And in cases where there’s a disruption in the model, such as items falling from a DC shelf, digital twins are updated in real time via sensors and AI to dynamically change these digital twins and deliver accurate information to other associated technologies, such as robots. This is where NVIDIA’s Mega comes into play, which is an Omniverse Blueprint for developing, testing and optimizing physical AI and robot fleets at scale in a digital twin before they are deployment in real-world facilities.
To go from idea to execution effectively, Martin noted that companies need to be aligned on a broad AI strategy. She offered the following tips:
- Have top executives act as sponsors;
- Identify key business challenges that AI could solve;
- Assign teams responsible for specific AI projects;
- Determine metrics that would help these teams measure the impact and value of these investments; and
- Start, measure and continuously provide updates on those metrics to ensure AI is adding value.