Walmart has put a precise number on what its generative AI shopping agent is worth: shoppers who engage with Sparky spend 35 percent more per order than those who do not, a figure CEO John Furner disclosed to investors in February during the company's fiscal fourth-quarter earnings call, turning what had been a promising experiment into one of the most concrete return-on-investment claims any major retailer has attached to an AI product.
The disclosure arrived alongside results showing Walmart's full-year revenues reached $713 billion, up 4.7 percent year over year. But the order-value figure stands apart from the broader revenue line because it speaks to something retailers have struggled to demonstrate since the generative AI wave began: that conversational AI can actually change what a customer puts in a cart, not just answer questions about return policies.
"The way we're using technology and AI is helping us create great customer solutions, reduce friction, simplify decision-making and pinpoint where our inventory is — all while maintaining the trust we've earned from our customers and members," Furner said on the earnings call.
What Sparky Actually Does
Sparky sits inside the Walmart mobile app and functions as a full-service shopping agent: it handles natural language queries, analyzes product reviews, surfaces recommendations, and completes checkout within Walmart's own system. About half of Walmart's app users have now interacted with Sparky, a scale of adoption that makes the 35 percent AOV lift more than an anecdote drawn from a thin test group.
Sparky is one of four "super agents" Walmart has built across its operations, alongside agents focused on supply chain, merchandise planning, and associate tooling. Furner framed the initiative in expansive terms on the earnings call: "Agentic commerce is going to be great for our customers. Agentic AI will be great for our associates, because it helps them focus on the things that are most important."
A separate Walmart survey cited by Adweek found that 81 percent of Sparky users have used the agent to look up product availability and details before purchasing — a behavioral pattern that explains the larger basket sizes. Customers who arrive at checkout having already compared options and confirmed availability through an AI-assisted conversation are completing a warmed purchase intent, not browsing cold.
The OpenAI Detour and What It Revealed
The success story has an instructive detour. From November 2025, Walmart made roughly 200,000 products available through OpenAI's Instant Checkout, allowing users to buy without leaving ChatGPT. The experiment ended in early 2026 after conversion rates through OpenAI's native checkout layer ran approximately three times lower than when customers clicked through to Walmart's own site. The fragmented experience — purchases processed separately from whatever sat in a customer's existing Walmart cart, often triggering multiple shipments — undercut the convenience Walmart's logistics infrastructure is designed to deliver.
Rather than exit AI-native platforms entirely, Walmart drew the opposite conclusion: embed Sparky inside ChatGPT and Google Gemini, but retain control of the cart and checkout. The ChatGPT integration launched in March 2026 for paid subscribers, with a free-tier rollout expected later this spring. A Gemini integration is planned to follow, and Walmart is in discussions with Anthropic about a potential deployment on Claude.
The pivot reflects a clear philosophy. Walmart is not trying to win inside third-party AI environments by ceding the transaction. It is trying to be where customers are already asking questions — and then routing them into the Walmart ecosystem for the close.
Ads Enter the Conversation
As Sparky's user base has grown, Walmart has begun layering in advertising. Brands can now buy sponsored placements within Sparky's responses, appearing as sponsored prompts when shoppers ask for product recommendations. Walmart began testing the format in fall 2025 and announced an expanded rollout in early 2026, positioning the inventory inside its Walmart Connect retail media network.
The move follows a year in which Walmart's global advertising revenue grew 53 percent year over year in fiscal Q3 2026. Embedding ads inside an AI agent that half of app users have already touched creates a high-intent inventory slot that few retail media competitors can replicate at comparable scale.
Automation Underneath the Intelligence
The Sparky story does not exist in isolation from the supply chain investment running below it. CFO John David Rainey told investors that roughly 60 percent of U.S. stores are now receiving freight from automated distribution centers, and that approximately 50 percent of e-commerce fulfillment center volume is automated. An AI agent that can recommend and complete a purchase only delivers on its promise if the item actually ships on time. Without the inventory-visibility and logistics automation Rainey described, Sparky's elevated basket sizes could translate into elevated disappointment rates.
The convergence of front-end AI and back-end automation is what Walmart frames internally as "AgenTek Commerce" — positioning Sparky not as a customer-service chatbot layered on top of an unchanged retail operation, but as the consumer-facing expression of a system transformation that reaches into warehouses and distribution logistics.
What to Watch
For the retail sector, Walmart's willingness to publish a specific lift metric raises the stakes on competitors who have deployed similar AI shopping tools without attaching numbers. If the 35 percent figure holds as the user base broadens beyond early adopters, it will become a benchmark against which Amazon's Rufus assistant and every other major retailer's generative AI investment will eventually be measured.
Furner's framing for the longer arc is worth holding: Walmart is pursuing AI through tech-company partnerships rather than pure internal build, letting partners develop the underlying capability while Walmart translates it into retail experience. That approach is pragmatic in the short term. Whether it proves defensible as AI platforms increasingly want their own share of the transaction is the more consequential question for the next twelve months.
"Agentic commerce is going to be great for our customers. Agentic AI will be great for our associates, because it helps them focus on the things that are most important."— John Furner, President and CEO, Walmart