Walmart

Philip is Director of Site Applied AI and Tools at Walmart, where he leads the development of AI-driven systems that support how teams shape the company’s digital storefront. With a background across banking, finance, and technology he brings a practical, operator-focused lens to applied AI in e-commerce.

In this feature, he reflects on evolving customer behavior, the realities of AI adoption, and how speed, data quality, and associate experience are shaping the next phase of e-commerce, while highlighting the need to balance innovation with practical usability at scale.

My path to leading applied AI at one of the world’s largest retailers did not begin in a server room. It began behind a bank counter. I managed branches, worked as a commercial loan officer, and spent years learning how decisions ripple through complex systems. That grounding in customer reality is precisely what informs how I think about technology today.

Tools Only Matter if they Actually Help People Do their Jobs Better.

After completing my MBA, I joined Walmart through its finance organization, moving across multiple functions before pivoting into technology roles that supported forecasting platforms and early machine learning tools. About a year and a half ago, I moved into e-commerce, where I now lead the applied AI and tools team responsible for the digital surfaces that millions of customers interact with every day.

My team sits at the intersection of associate experience and customer experience. If you land on Walmart.com and notice a seasonal banner for Mother’s Day or a carefully curated product layout, there is a dedicated team deciding which items to feature, how the page should look and feel, and what message to deliver. We build and support the tools they use to manage that content. What makes this role distinctive is not just the technology being deployed, but the discipline required to ensure technology is utilized to improve productivity.

The 40 Percent Problem

When the conversation turns to scaling applied AI in e-commerce, most discussions begin and end with data quality or model deployment. I see it differently. There are two distinct scaling challenges operating simultaneously, and both demand equal attention.

On the customer side, shopping behavior is evolving in unpredictable ways. Generative AI tools like ChatGPT have only been part of mainstream consumer life for roughly 6 to 12 months. Customers are still working out how to shop with AI assistance, what to expect from it and how to trust it. That uncertainty makes it difficult to design a fixed experience. Our responsibility is to remain agile, monitor behavioral data closely, and continuously adapt the experience as understanding deepens.

On the associate side, the barrier is more psychological than technical. If an AI tool handles 60 percent of a task and leaves 40 percent for the person to finish, it can feel more burdensome than simply doing the task manually. The perception that AI adds work rather than removes it is, in my view, the single most damaging belief the industry must overcome. We are tackling it by improving explainability and transparency, making the reasoning behind model recommendations visible and legible, so that associates understand exactly why the AI is suggesting what it is suggesting. As these insights become clearer, adoption accelerates.

“AI Only Works When it Truly Reduces Effort. If It Leaves Meaningful Work Behind, People Will Struggle to Trust And Adopt it.”

From Static Pages to Generated Experiences

Most E-commerce platforms today are largely static pages. A page loads, some elements are personalized based on prior behavior and the experience is delivered. It remains functional, but not moment-responsive. We are actively building towards real-time, generated experiences where the entire layout of a page is tailored for a specific customer the instant they arrive.

To make this tangible, consider the challenge we face with product discovery at Walmart. With one of the largest SKU counts in retail and a third-party marketplace that expands the assortment even further, the scale is massive. Identifying which items to feature on site has historically been a largely manual and time-intensive process.

We have been rolling out AI tools designed to accelerate that discovery, surfacing more relevant candidates faster, improving accuracy, and ensuring that featured products are accurate and locally in stock. Associates save significant time, while customers benefit from better, more personalized recommendations, with fewer friction points between discovery and purchase.

The shift toward fully generated experiences carries two tensions worth managing carefully. The first is the balance between deep personalization and broader business objectives, between showing a customer what they already love and introducing them to something strategically important but unfamiliar. The second is data integrity. These systems depend entirely on the signals they receive, and a single low-quality data point can undermine an entire model.

Companies that have built strong data foundations and a genuine understanding of their customers will hold the lasting advantage as this space matures.

The Associate of The Future as a Strategic North Star

My leadership approach is less about managing projects and more about maintaining clarity of direction under conditions of constant change. At the center of that framework is a single question I return to consistently: What do I want the associate experience to look like three or four years from now?

That future-state picture functions as a north star. From that clear vision, I work backward to identify problems standing between today’s reality and that vision, explain why each one matters, and then prioritize the problems that will generate the greatest impact for customers and associates.

This framework creates natural alignment. When that line of reasoning is made explicit, I find that resistance from product, engineering, and business stakeholders largely disappears. The conversation shifts from “should we do this?” to “which of these do we tackle first?”, a far more productive place to operate from.

The Coming Speed Revolution in E-Commerce

I believe the next significant unlock will come from speed. The models available today are already reasonably capable of identifying what a customer might want. The remaining constraint is processing that insight at the individual level in milliseconds.

When a model can understand a customer’s intent the moment they land on a page and respond before they consciously register having arrived, the nature of e-commerce will change in ways that are currently difficult to fully imagine. The technical barriers, such as bandwidth, capacity and processing speed, are current but narrowing, but the question is how prepared our teams and tools will be when new possibilities arrive for e-commerce expansion.

For those building careers in this space, my counsel is straightforward. Domain knowledge helps, but adaptability is the non-negotiable. Relationships matter more than technical credentials. And the ability to update your approach as new information arrives, without losing sight of your goal, is what separates people who simply accumulate experience from those who genuinely accelerate in their growth.

My own path from banking to leading AI tools at Walmart has reinforced one consistent truth: technology moves fast, but lasting progress comes from keeping people at the center. By focusing on the associate experience, maintaining clarity on the problems we solve, and staying agile with our customers, we can turn the promise of AI into real, everyday value on one of the world’s largest e-commerce platforms.