Over the next few years, Walmart is aiming to equip each of its 2.1 million employees — from its tech team to its signature in-store greeters — with some level of AI skills, EVP and Chief People Officer Donna Morris said at MIT Technology Review’s EmTech AI summit Tuesday.

Morris described the business as “people-led, tech powered,” saying the company has been experimenting with AI and honing a strategy since the rise of generative AI in the fall of 2022. 

By the following year, Walmart had built an internal platform for associates to experiment with AI. The company has since streamlined its portfolio, standing up four different agent platforms with a mix of custom-built and external large language models. The company has a more than 35,000 person tech team, but AI touches every part of the organization, Morris said. 

Customers may be familiar with Sparky, the company's personalized recommendation platform. But executives have a big focus on its internal AI platform, Morris said. Walmart partners with OpenAI and Google Gemini on role-specific AI certifications offered through the company’s associate-facing platform, Squiggly. 

The certifications are currently open to associates in the U.S. and Canada, 1.7 million of the company’s workforce. The training aims to bring agentic AI into associates’ workflow where it removes the most amount of friction for them. 

For a front-of-house worker, that could mean more quickly finding items in a store’s stock room, or having an AI agent help translate a conversation with a customer who speaks a different language, Morris said. 

“We need to bring people along on the journey, and we need to make sure that we're educating people,” Morris said. “It's a form of literacy in many ways, and so we really want to make sure that our associates feel literate and capable of using AI.”

Data and ROI

When building their AI stack, Morris said tech leaders considered what data was already publicly available. If something exists in the public domain that's beneficial, they’ll use an external LLM to drive the data. But if it's specific to Walmart, they’ll use their internal data lake.

“This actually is not too different from how we worked in a pre-AI world,” Morris said. “There was certain data and certain information that was only accessed based on positions and decision-based rules. That's the same type of framing today, but it's in an agentic world.”

As companies review their investment priorities around automation, Morris said Walmart’s perspective on ROI may be different from other enterprises. The goal of AI integration, especially for their customer-facing roles, is creating more time for face-to-face, customer-centered work. 

The retail sector is moving quickly to deploy AI tools in pursuit of enhanced customer experiences and lower costs, with more than eight in 10 retailers integrating AI into their operations as of late last year. Retailers reported using AI for marketing, IT and digital functions, commerce, merchandising strategy and pricing, a BRG report found. 

“If we can actually provide technology that allows associates to serve customers more effectively, we'll say that's value added,” she said. 

Morris said the company hasn’t experienced role displacement because of their AI use. It has the same amount of employees as it did when she joined the organization six years ago, she said, but they’re seeing much higher revenue. 

In recent years, the company has also adjusted its business model to include digital advertising, data ventures and a healthcare business. 

AI is a part of Walmart’s world, but it’s not the whole world, Morris said. 

“The vast majority of our jobs will have some shaping. They won't have fundamental reshaping,” she said. “All of us have to really make sure that we're pragmatic about how much AI is truly shaping and what jobs.”

Correction: This article has been updated to reflect Berkeley Research Group is now known as BRG.