Walmart spent much of the past year describing Sparky, its generative AI shopping assistant, as a better way to search. As the company heads into its June 4 annual shareholders’ meeting, it is describing the tool differently, presenting Sparky as software that anticipates what a customer needs and acts on it, proposing dishes to cook, assembling the ingredients into a cart, and prompting reorders before the shopper asks. In his first letter to shareholders as chief executive, filed alongside Walmart’s annual report in April, John Furner highlighted investments in AI-powered solutions to serve customers, and Sparky is the most visible of them.
Walmart launched Sparky in June 2025, putting it behind the smiling “Ask Sparky” button in its mobile app to find items, make sense of reviews, and prepare for occasions like a game night or a beach trip, according to the company’s own announcement. A year on, the assistant Walmart describes is more active. Daniel Danker, Walmart’s executive vice president of AI acceleration, product and design, said earlier this year that Sparky can recognize when a customer buys the same items week over week and prompt the next order on its own, according to Talk Business & Politics. The same report described the assistant surfacing reminders for things like an auto-service appointment or a prescription ready for pickup.
The figures Walmart is putting behind the tool are what move it from promotion to something suppliers should track. Furner told investors on the company’s fourth-quarter earnings call that customers who use Sparky carry an average order value roughly 35% higher than those who do not. David Guggina, named president and CEO of Walmart U.S. in February, said about half of the retailer’s app users have tried the assistant. A Walmart survey found that 81% of shoppers who have used Sparky did so to check availability or review product details before buying. Guggina put the commercial logic plainly, saying Sparky is helping Walmart “evolve from traditional search to intent-driven commerce.”
The distinction matters for anyone who plans around how Walmart shoppers find products. A search box rewards the supplier whose listing matches the query. An assistant that proposes the dinner and then fills the cart rewards the supplier whose product the assistant chooses to include. Walmart has said it expects a conversational, multimodal interface to take over much of the work the search bar does today, and it has pointed to capabilities like reading a photo of a shopper’s pantry to suggest recipes as where Sparky is heading. Those are described as the destination, not the current state: an early hands-on review by Simply Recipes at Sparky’s 2025 launch found its meal-planning and party-planning results uneven. Suppliers should treat the anticipatory framing as direction of travel rather than a finished system, and build for that direction now.
The dish-suggestion capability lands hardest in grocery, Walmart’s largest business and the category where it most wants to grow online orders. When a shopper asks Sparky what to make for dinner and accepts a recipe, the assistant decides which products become the ingredients, and that selection is the moment of placement. It does not behave like a search-results page. The implication for grocery suppliers is that the inputs an assistant leans on to choose confidently, meaning complete and accurate item content, clear attributes, and healthy review signals, carry more weight when the shopper is asking for an answer than when they are scrolling a list. A thin or ambiguous listing that still surfaces in a keyword search can be passed over by an assistant assembling a meal.
Where discovery shifts, monetization tends to follow, and Walmart has already begun building the advertising layer into its agents. Walmart Connect’s Marty, the company’s agent for suppliers, sellers and advertisers, is moving from beta toward all sponsored search advertisers in the first half of 2026, and Walmart has said it is now running ad formats inside Sparky itself after a pilot.
How that plays out depends on the seller. For 1P suppliers, the levers are catalog and item-setup quality, plus the Walmart Connect programs now reaching into conversational surfaces, since those are the inputs and the paid placements an assistant draws on. For 3P Marketplace sellers, the same content and review signals apply, but listing completeness and quality scores carry more of the load, because a Marketplace listing an assistant judges incomplete is less likely to be the one it recommends. Both segments are exposed to the same shift, but the path by which a given item earns a recommendation is not the same for each.
The reach of all this extends past Walmart.com and the app. Walmart announced a shopping experience inside OpenAI’s ChatGPT in October 2025 and one pairing its assortment with Google’s Gemini in January 2026, extending AI-assisted discovery of its products onto surfaces it does not own. Walmart has said it wants online sales to reach roughly half of total revenue within about five years, up from close to a fifth, and the conversational interface is the bet meant to get shoppers there. For suppliers and sellers, that target is the context that makes the assistant worth planning around now: the surface that decides what a Walmart customer sees is being rebuilt around answering the shopper rather than listing options for them.