Walmart’s Enterprise AI Blueprint: Trust Engineering at Scale

Walmart We continue to crack code by deploying agent AI at an enterprise scale. Their secret? Treating trust as an engineering requirement is not the last compliance checkbox to check.

During the “Life in Algorithms: How Walmart Agent AI Redefines Consumer Trust and Retail Leadership” session VB conversion 2025, New Technology Desirée Gosby’s Vice President of Walmart, How Retail Major Operates thousands of AI use cases. One of the retailers’ main objectives is to consistently maintain and strengthen customer trust among the 255 million shoppers.

“We think this is a pretty big inflection point that’s very similar to the internet,” Gosby told industry analyst Susan Etlinger during a session Tuesday morning. “It’s profound in terms of how we actually work and how we actually work.”

This session provided valuable lessons learned from Walmart’s AI deployment experience. Implicitly throughout the discussion, we continue to search for new ways to apply the principles of distributed systems architecture, avoiding the creation of technical debt.

>> See all 2025 coverage here <

Four Stakeholder Frameworks constitute AI deployment

Walmart’s AI architecture rejects a horizontal platform for targeted stakeholder solutions. Each group receives a dedicated tool to deal with specific operating friction.

Customers are involved Sparky For natural language shopping. Field Associates get inventory and workflow optimization tools. Merchants access decision support systems for category management. Sellers receive business integration capabilities. “And of course we have developers. You really, you know, give them superpowers and charge them with new agents of tools,” Gosby explained.

“We have thousands, if not thousands, use cases across the company we bring to life,” Gosby revealed. Scale requires the construction discipline that most companies lack.

This segmentation allows the basic needs of each Walmart team to install dedicated tools for specific jobs. Inventory management associates need a different tool than merchants analyzing local trends. Common platforms fail because they ignore operational reality. Walmart’s peculiarity promotes recruitment through relevance rather than mission.

Trust Economics is promoting AI adoption at Walmart

Walmart has discovered that trust is built through value delivery, not only through valuable training programs, but also through value delivery.

The Gosby example resonated as he explained the evolution of mothers’ shopping from weekly store visits to delivery during the Covid era, and explained how it accurately demonstrates the natural adoption mechanism. Each step provided immediate and concrete benefits. There was no friction and no forced change control, but progression happened faster than anyone could have predicted.

“She’s been interacting with AI all the time,” Gosby explained. “The fact that she could go to the store and get what she wanted was on the shelf. AI was used to do that.”

The benefits customers get from Walmart’s predictive commerce vision is further reflected in Gosby’s mother’s experience. “Instead of having to go every week, find the groceries you need to deliver, that’s the essence of predictive commerce, and that’s how it delivers massive value to all Walmart customers.

“If you add value to their lives, help them remove friction, help them save money, which is part of our mission, and live better, then trust comes from it,” Gosby said. Associates follow the same pattern. Adoptions happen naturally and earn trust when AI actually improves their work, saves time and helps them to be superior.

The fashion cycle is compressed over months to weeks

Walmart’s trends towards product systems quantify the operational value of AI. The platform integrates social media signals, customer behavior, and local patterns to reduce product development over months to weeks.

“The trend towards products has kept us from months to weeks to weeks to reach our customers with the right product,” Gosby revealed. This system creates products based on real-time demand, not historical data.

A few months of compression will change Walmart’s retail economics. Inventory returns accelerate. Reduces markdown exposure. Increased capital efficiency. The company maintains its pricing leadership while matching its competitors’ capabilities to the market. All fast categories can benefit by using AI to reduce time to market and providing quantifiable profits.

How Walmart uses the MCP protocol to create a scalable agent architecture

Walmart’s approach to agent orchestration draws directly from the hard experiences of distributed systems. The company uses Model Context Protocol (MCP) to standardize how agents interact with existing services.

“We’re really looking at how we break down the domains and wrap them up as MCP protocols, and we’re going to expose what we can start tuning different agents,” Gosby explained. This strategy transforms existing infrastructure rather than replacing it.

Architectural philosophy runs deeper than protocols. “The changes we saw today are very similar to what we saw when we went from the monolith to the distributed system. We don’t want to repeat these mistakes,” Gosby said.

Gosby outlined the implementation requirements. “How do you break down your domain? Which MCP servers should you have? What agent orchestrations do you have?” At Walmart, these are not theoretical exercises, but rather daily operational decisions.

“We want to take existing infrastructure, break it down and reconfigure it into agents who want to build it,” Gosby explained. This standardized first approach allows for flexibility. A service built many years ago is a power agent experience through the right layer of abstraction.

Merchant’s expertise becomes Enterprise Intelligence

Walmart leverages decades of employee knowledge and is a central component of growing AI capabilities. The company systematically captures category expertise from thousands of merchants, creating a competitive advantage that digital-first retailers cannot match.

“We have thousands of merchants who are great at what they do. They are experts in the categories they support,” Gosby explained. “We have cheese merchants who know exactly what wine is and what cheese pairings they have, but that data is not necessarily captured in a structured way.”

AI manipulates this knowledge. “We can use the tools we have to capture the expertise they have and really stand that to our customers,” Gosby said. The application is specific. “When they’re trying to figure it out, hey, I need to throw a party, what kind of appetizer should I have?”

Strategic advantage compounds. Decades of merchant expertise will become accessible through natural language queries. Digital-first retailers lack this foundation of human knowledge. Walmart’s 2.2 million associates represent unique intelligence that algorithms cannot synthesize independently.

New metrics measure autonomous success

The Walmart Pioneers measurement system is designed for autonomous AI rather than for human-driven processes. Traditional funnel metrics fail when an agent handles an end-to-end workflow.

“In the agent world, we’re beginning to get through this and that’s going to change,” Gosby said. “The indicators about transformations and things like that will not change, but we will consider completing the target.”

Shifts reflect operational reality. “Did we actually achieve it? What is the ultimate goal that our associates, customers are actually solving?” Gosby asked. The question reconstructs the success measure.

“After all, is it a measure, and we are offering benefits? Are we offering the value we expect, and then go back from there and get the correct metrics basically?” Gosby explained. Problem solving is more important than process compliance. How AI helps customers achieve their goals comes first over the conversion funnel.

Enterprise Lessons from Walmart’s AI Transformation

Walmart’s Transformation 2025 session provides actionable intelligence for enterprise AI deployments. The company’s operational approach provides a framework that is validated at scale.

• Building discipline will be applied from the first day. With the transition from monolithic to distributed systems, Walmart has provided the lessons they need to learn to succeed in AI deployments. An important lesson learned is to build a good foundation before expanding and defining a systematic approach to prevent expensive rework.
• Match your solutions to the needs of your specific users. One-Size-Fits-AI AI fails every time. Store Associates need a different tool than merchants. Suppliers need different features than developers. Walmart’s target approach drives adoption.
• Build trust through proven values. Start with a clear victory that provides measurable results. Walmart has moved step by step from basic inventory management to predictive commerce. Each success gains the following insights and knowledge:
• Turn employee knowledge into enterprise assets. There are decades of specialized expertise within the organization. Walmart systematically captures merchant intelligence and operates operations across 255 million weekly transactions. This institutional knowledge creates a competitive advantage that algorithms cannot replicate from scratch.
• Measure what is important in an autonomous system. Conversion rates miss the point where AI handles the entire workflow. Focus on problem solving and value delivery. Walmart’s metrics have evolved to match operational reality.
• Standardize complexity before it hits. The integration failure killed more projects than bad code. Walmart’s protocol decisions prevent the confusion that derails most AI initiatives. The structure enables speed.

“It’s always coming back to basics,” Gosby advised. “Take a step back and understand the problems you really need to solve for our customers, and for our peers. Where is the friction? Where is the manual job where you can start thinking differently now?”

Walmart’s blueprint expands beyond retail

Walmart shows how enterprise AI can succeed through engineering discipline and systematic deployment. The company handles millions of daily transactions at 4,700 stores by treating each stakeholder group as a clear challenge requiring tailored real-time solutions.

“It’s permeating everything we do,” Gosby explained. “But at the end of the day, the way we see it is that we always start with our customers and members and really understand how it affects them.”

Those frameworks apply to the industry as a whole. Financial services organizations balance customer needs and regulatory requirements, healthcare systems that coordinate patient care across healthcare systems, providers, and complex supply chain managers all face similar multi-stakeholder challenges. Walmart’s approach provides a tested methodology to address this complexity.

“Our customers are trying to solve their own problems, and it’s the same for their peers,” Gosby said. “Did we actually solve that problem with these new tools?” This focuses on problem resolution rather than technology deployment, facilitating measurable results. Walmart’s scale validates the enterprise’s approach that is ready to move beyond pilot programs.