Walmart earns around USD 570 billion annually, making it the world’s largest company in terms of revenue. The retail giant operates more than 10,000 stores across 24 countries across the globe. Even though Walmart does not operate any direct-to-consumer physical stores in India, a major part of its tech team is based in the country.

Recently, Walmart CTO Suresh Kumar said that India is Walmart’s tech backbone and its Global Tech centre located in India is even bigger than that in the US.

At the recently held AI@Walmart event at Walmart’s Global Tech centre in Bengaluru, AIM caught up with Balu Chaturvedula, vice president-technology, supply chain, US omni tech. Balu has over 29 years of experience in the industry. He began his career at Walmart Global Tech in 2015 as a senior director. At Walmart, Balu and his team are building a next-gen e-supply chain management solution. 

“We are trying to create the next-generation retail business. Our focus on automation is big and we’re actually not just looking at only logistics and distribution. We are actually trying to look at what the next-generation Walmart store is going to look like?” Balu said.

In this exclusive interaction, he further discusses how the retail giant has invested heavily in automation, AI-driven algorithms, and other cutting-edge technologies to streamline its supply chain, optimise its inventory management, and provide personalised customer experiences. 

Leveraging  generative AI

In recent months, generative AI has gained significant momentum and presents numerous potential use cases for retailers. Balu believes the next big disruption in generative AI is going to be in the retail industry as it will change the way customers shop.

In fact, long before ChatGPT came about, Walmart has been building its own natural language understanding capability called Converse. By leveraging the GPT-3 large language model by OpenAI, and building a layer around it, Walmart has managed to customise and adapt it for the Walmart environment.

Currently, a decent part of Walmart’s customers’ responses are handled by its internal chatbot without any interventions. Further, Walmart has also launched AskSam, which Walmart associates can use to get information about anything, saving them time and increasing productivity. The NLP of AskSam too is done by Converse.

In January this year, Walmart launched text-to-shop, a feature that enables customers to use text messages to order products for pickup or delivery.

Now, Walmart is contemplating leveraging GPT-4, the most-advanced large language model so far, to enhance its text-to-shop feature. Further, since GPT-4 is multimodal, pretty soon it will be able to process images as well. This is going to be significant for Walmart to offer a better shopping experience to its customers.

Balu also believes generative AI has a wide range of use cases for Walmart, such as improving internal processes or productivity.  For example, the temperature in the grocery distribution centre could range from -18 degrees to -26 degrees, whereas the freezer section runs on a temperature as low as -40 degrees.

For Walmart associates, working in such conditions sometimes is challenging. “So I think generative AI can be so powerful if you build that cognitive learning, and then either through vision picking or voice picking, we might be able to ensure that the associates don’t have to get out of the car. You can actually give instructions and then forklifts can work, bring and then keep the requisite items. 

“However, I think we’re still in the nascent stage and we are still figuring out the right balance between leveraging what’s out there and leveraging what we have internally,” he said.

Automation

As Walmart integrates automation and expands into higher-margin businesses such as advertising, last-mile delivery, and fulfilment services, CEO Doug McMillon anticipates that the company’s profits will increase at a faster rate than its sales in the next five years.

With regard to automation, Balu said that Walmart has launched its first state-of-the-art fully automated fulfilment centre in Chicago.

“I think we are looking at 5-6x more productivity, and probably another 10x more accuracy in terms of either pick walks or inventory management,” Balu said.

In this light, last year in October, Walmart acquired e-grocery automation firm Alert Innovation. The retail giant was drawn to the custom-built inventory-handling technology developed by Alert Innovation that employs robots capable of omnidirectional movement without the need for lifts or conveyors, enabling the efficient storage, retrieval, and dispensation of orders.

Walmart is also looking to add more fully automated fulfilment centres and the next one is coming up in Indiana in the next two months or so. However, Balu revealed that automation does not always mean efficiency.

“When we are looking at the entire processes, what are the two or three important aspects that are inefficient? And how do you actually use machine intelligence to make that smarter and better? I think we actually have mixed modes. We have buildings that completely run on machine intelligence, and in some buildings, only certain processes run on machine intelligence,” he said.

A greater focus on last-mile

Over the years, Walmart has also invested heavily in improving its last mile delivery system. “We’ve launched our own last mile pickup and delivery app. We’ve massively upgraded our delivery speeds from five to seven business days to a day or two,” said Balu.

He added that Walmart has also experimented with drone delivery in many cities across the US. Currently, Walmart operates 36 drone delivery hubs across seven states, including Arizona, Arkansas, Florida, North Carolina, Texas, Utah and Virginia.

(Source: Walmart)

Till date, the retail giant has safely completed more than 6,000 drone deliveries to customers in as little as 30 minutes. For drone delivery, Walmart has partnered with DroneUp, Flytrex and Zipline.