Amazon and Walmart Push AI Agent Shopping from Product Selection Toward Checkout

Walmart and Amazon this week advanced AI agents on the consumer and enterprise procurement sides respectively, bringing agents closer to the point where transactions actually occur. According to a Walmart announcement on October 30, its shopping agent Sparky has added several features: users can upload a photo of a handwritten shopping list or import a digital list to generate a cart directly, and can also upload photos of problems such as tablecloth stains to receive product recommendations; Sparky can also generate recipes based on budget and dietary needs, and add ingredients to the cart after user confirmation.

Walmart is also connecting its online assistant to physical stores: the Shop to Light feature allows customers to locate products on a store map and light up electronic shelf labels. Walmart said voice functionality is expected to launch in all stores by the end of the year, separate from features already available.

Amazon, meanwhile, discussed how much purchasing authority enterprises should give agents at the Amazon Business conference held in Nashville in October. According to a November 4 report by Procurement Magazine, Céline Vuillequez, Vice President of Amazon Business Europe, proposed four points: setting spending limits, embedding company policies into the system, requiring human approval when limits are exceeded, and fully recording agent actions.

In the same report, Matt Yanchyshyn, Vice President of AWS Marketplace and Partner Services, said the agent model on AWS Marketplace raised conversion rates from about 3% to 6%, a figure specific to AWS Marketplace. He also expects agent payments to scale by 2027, including routine renewals without human intervention, but this remains a prediction.

Source: https://www.pymnts.com/news/artificial-intelligence/2026/amazon-walmart-bring-ai-agents-closer-purchase/