Two of the largest retailers in the United States shipped major upgrades to their AI shopping assistants within six days of each other, and the timing is not a coincidence. Both are trying to own the moment when a vague intention becomes a filled cart.
Amazon said on October 29 that Alexa for Shopping can now handle more than 100 everyday shopping actions across its store. The list runs from creating and editing lists and registries to managing Subscribe and Save deliveries, changing delivery instructions, resolving billing questions, checking gift card balances and tracking refunds. The assistant is available in the Amazon Shopping app and on the website, and the company says more than 400 million customers used it in the past twelve months, with the number using its agentic features up more than eight times since the start of the year.
Rajiv Mehta, Amazon vice president of conversational shopping, described the intent as making every part of shopping as convenient as possible, and said the company is continuing to invent new ways to save customers time. That is corporate language for a specific ambition: shift the default interface from search and browse to a conversation that ends in a purchase.
Walmart answered on October 30 with six new capabilities in Sparky, its own AI shopping agent. Shoppers can now photograph a problem such as a stain on a sweater and get a product recommendation, or upload a picture of a handwritten list and have it converted into a cart. A feature called Shop to Light pins an item on a store map and then flashes the electronic shelf label once the shopper reaches the right aisle. Sparky can also generate recipes and load the ingredients, suggest healthier substitutions in a recipe or an existing cart, and converse in Spanish and French. Voice is expected across the chain by the end of the year, and the assistant is scaling into Canada, Mexico and Sam Club members in the United States.
Walmart has been unusually willing to publish the commercial effect. Chief executive John Furner said in August that usage of Sparky has risen 70 percent since its June 2025 debut and that customers who use it spend 40 percent more per order than those who do not. US chief executive David Guggina said the introduction of visual search produced a 57 percent lift in conversion compared with text-based searches.
Those numbers explain why both companies are investing here rather than in another recommendation widget. An assistant that completes tasks is not a discovery layer. It is a checkout layer with memory, and whoever owns it can influence which brand fills a line that says paper plates.
That is the part brands should read twice. Walmart's own example, published with the feature, shows Sparky telling a shopper that a previously purchased trash bag brand is out of stock at her usual store and then recommending alternatives for items she has not bought before, producing a nine-item cart of about $64. The mechanism is revealing: purchase history determines the default match, store-level inventory determines availability, and where neither exists the assistant chooses. In a cart built from a handwritten list, the shopper never typed a brand name, so the selection happens inside the retailer's system.
Amazon is building the same gravitational pull from a different direction. Alexa for Shopping can set a price alert and then buy the item when it hits the target, assemble a cart from past orders, and compare products side by side from search results. Its Buy for Me function completes purchases from other retailers on the shopper's behalf. Each of those features moves a transaction closer to a place where Amazon sees the intent, the price history and the payment instrument at the same time.
The competitive field is filling in around them. Target is beta testing an assistant called Ask Target on its site and app, and Best Buy launched Ask Blue to give expert advice on product selection. Each of these is a defensive move as much as an offensive one: if a third-party agent becomes the customer default interface, the retailer becomes fulfillment infrastructure and loses control of the shelf.
Walmart has tried to get ahead of the obvious objection. Furner published an open letter stating that engaging with Sparky is an invitation to serve customers better, not to use personal information to set a personalized price, and that the company has never used associate relationships with customers to charge more and will not do so with AI. The denial is notable because Walmart's digital shelf labels have been at the center of dynamic pricing questions, and because an assistant that knows a shopper budget is exactly the system capable of exploiting it.
For suppliers, the operational consequences are already concrete. Product data quality stops being a marketing concern and becomes a distribution concern: if an attribute is missing or a description is ambiguous, an agent has less to work with and defaults to something it has seen before. Purchase history becomes a moat that new brands cannot buy their way past with a sponsored placement, at least until the assistants disclose how recommendations are ranked and whether placement can be purchased.
Neither company has published that ranking logic. Until they do, the honest position for a brand is to assume two things at once: that first-party purchase history will decide most default matches, and that structured, complete and accurate product records are the only reliable way to be the option an assistant can justify. The cart conversation is now a channel, and channels get optimized for whether or not anyone announces the rules.
