AI Shopping Agents Are Growing Fast. Most Shoppers Still Aren't Sold.
Barron's reports that Muse passed five million downloads in just over three weeks, while Bain estimates agentic shopping could reach $500 billion within four years, but a UBS survey finds only 17 percent of Americans think agentic AI is a good idea.
By Icarus4 min read
Meta’s Muse passed five million downloads in just over three weeks after its September launch, topping both the Apple App Store and Google Play. Bain & Company estimates that agentic AI could account for as much as $500 billion in spending within four years, or as much as a quarter of all US e-commerce by 2030. And yet most shoppers are not convinced: a September UBS survey found only 17 percent of respondents thought agentic AI was a good idea, while 43 percent said it was not. A new Barron’s report on the rise of AI shoppers examines that gap between adoption and trust.
Support slipped from the prior quarter. The share calling agentic AI a good idea fell a percentage point, while the share calling it a bad idea rose a point. Just over 10 percent said they would use agentic AI to book a restaurant, down from about 12 percent in June. “Perhaps six or 12 months ago, we would have expected steady growth in the percentage of people becoming aware of and using AI,” UBS analyst Jay Sole wrote. “The September data indicates that is not exactly playing out, at least for now.”
The Trust Problem
Barron’s traces the skepticism to a basic question shoppers cannot answer: how does the agent pick what it picks? Columbia Business School professor Omar Besbes, whose research Barron’s cites, found that different agents return very different recommendations for the same query, even for simple products like toothpaste. If a bot picks an item that fits a budget and a set of criteria, there is still no way to know whether better options existed, or why the agent chose the one it did.
Users also cannot see the agent’s reasoning, and there is no uniform legal standard for who is responsible if something goes wrong, such as fraud. “These agents don’t work for you,” Lindsay Owens, president and CEO of nonprofit watchdog Groundwork Collaborative, told Barron’s. “We could be unleashing the most sophisticated snake oil salesman the world has ever seen.”
Meta, asked for comment, pointed Barron’s to publicly available statements about Muse’s safeguards and partnerships for securing transactions. OpenAI did not respond.
Big Retailers Are the Early Winners
For now, the agents’ recommendations tilt toward the biggest names. Companies like Walmart, Target, and Costco typically take the top spots in agent recommendations, along with category specialists such as Home Depot and Best Buy, which the agents read as reliable and well stocked. UBS analyst Michael Lasser’s research found that AI agents favor traditional retailers over digitally native ones.
Retailers now compete for bot recommendations as well as shelf space, app downloads, and search rankings, Lasser said. Retailers that feed agents structured catalogs, inventory data, reviews, and store information may gain an advantage. The old levers may work differently. Besbes found that bots tend to avoid products labeled “sponsored” in favor of those deemed the “overall pick,” even in simulations where the labels did not correspond to real rankings.
What It Means for Muse
Muse is at the center of this tension. Shopping is its most common use, according to Meta’s third-quarter earnings call, with more than a quarter of users shopping through the agent and average purchases above $200. Meta has set up checkout through Stripe’s Link and Shop Pay, with protections including free coverage for damaged items and no-fee returns. Amazon blocked the agent from shopping on its platform in September while Walmart opened its own agent gateway the same month.
The piece also argues that the agents’ rise may reorder who benefits from e-commerce. “This could be an iPod moment for AI, when a breakthrough product makes a developing technology available to the mass audience,” David Russell, TradeStation’s global head of market strategy, told Barron’s. Trust is still low and recommendations still favor incumbents, so the winners are far from settled. As Guggenheim analyst Simeon Siegel put it, “Anyone writing a list of winners now better be using a pencil.”
