This new AI tool helps Walmart’s merchandising team plan what’s in stores

Within Walmart, employees known as merchants make decisions about which products the company carries online and in stores, as well as pricing for those items.

Naturally, the job involves plenty of data analysis, with merchants breaking down sales numbers by product category, sales channel, region, brand, item characteristics, and other factors. But manually running all of those reports and examining results using tools like Excel can be time-consuming, especially when merchants need to run multiple reports.

“You can see how these reports can become time-consuming when analyzing customer behavior across so much data,” says Brian Knapp, senior vice president for merchandising transformation at Walmart U.S. “In fact, running and analyzing these reports can take hours, and we know that customer expectations are changing fast, and we need to be able to move with speed to respond to customer demand.”

To help merchants quickly access and analyze relevant information, Walmart has introduced Wally, an internal generative AI tool that can dive into internal data to create responses in just seconds. Wally uses a familiar chat-style interface to retrieve relevant data from Walmart’s databases while accurately interpreting product industry jargon and category names (meaning users don’t have to worry about whether the database lists an item as, say, “TV” or “television”). Wally can then generate quick answers, tables, or full reports as needed.

“We’re able to take some of the best practices we as a company have [learned] over the decades in analyzing this kind of data and generate a very high quality report for our merchants to then go use,” says Aditya Kumarakrishnan, a distinguished data architect at Walmart Global Tech.

In a demo, Kumarakrishnan showed how a merchant could use Wally to analyze data about consumer electronics sales, probing details about televisions brands and prices for both in-store and delivery purchases. Wally can also help merchants manage both out-of-stock or overstocked items, and determine when to adjust prices or update sale info.

The AI tool, first announced to the public on Thursday, was introduced to merchants about a month ago. So far, Knapp says, they’ve reported it has saved time they can then use on other tasks, and has made data more accessible to merchants. Wally has also been given access to training material that Walmart provides to its merchandising team, allowing it to make recommendations and interpret data in line with that internal guidance.

Wally isn’t the only AI tool in use at Walmart. The company has previously unveiled an AI shopping assistant for customers and highlighted generative AI features to guide web customers to relevant deals. In an earnings conference call Thursday, CEO Doug McMillon said AI coding assistance and code completion tools helped save about four million developer hours last year. 

To make sure Wally gives accurate answers, Walmart has developed automated tests where the tool’s numeric responses are checked against known answers; the company even trained an AI judge to evaluate the software’s conversations based on human-annotated samples. Kumarakrishnan says his team also regularly reviews feedback from users on how the tool is working and other features it could offer.

“There’s a whole group of us that love sifting through the feedback that merchants are giving,” he says. “There are questions that they’ll ask and they’ll want to know a slightly different angle, and that gives us a roadmap to go execute on for Wally.”

View the full article