Your next personal shopper may not be a person at all. The latest breed of artificial intelligence assistants goes beyond the abilities of chatbots, which are great at answering questions and creating content. These generative AI chatbots, like ChatGPT, are the brain. These new AI agents layered on top, think of them like the body , can now autonomously perform tasks for users, including shopping. They are also persistent, always running in the cloud. Meta Platforms ‚ new personal agent Muse, released Sept. 8, can search retail sites for the best products, compare prices, and even complete purchases. It wasn’t the first. But it quickly went viral, displacing ChatGPT at the top of Apple ’s U.S. App Store charts. Muse has surpassed 5 million downloads, with more than 3 million weekly active users. Meta shares have surged more than 17% since the Muse launch. This evolution from generative AI (which creates something for you) to agentic AI (an agent that does something for you) raises questions for many companies and their investors. Chief among them: What happens when consumers hand over purchasing decisions to AI agents? The early read on Wall Street is that some companies and industries face steeper challenges than others — and some may actually benefit from the shift. For example, companies that benefit from customers not bothering to cancel subscriptions, compare prices, or search for better deals have taken a beating in recent weeks. Since the launch of Muse through Thursday’s close, Planet Fitness fell almost 12%, Booking Holdings dropped 11%, SiriusXM dropped 8.6%, and Charles Schwab lost 9% — all victims of the so-called “ consumer inertia “ trade. Goldman Sachs described the potential transition as moving from „search and shop“ to „delegate and approve.“ Today’s AI shopping tools are still largely used for product discovery and recommendations, but Goldman sees them eventually executing more of the purchasing process. So far, however, investors don’t appear to be pricing in much near-term disruption for our three retail holdings. Since Muse’s launch through Thursday’s close, TJX has gained nearly 8%, Costco is up about 4%, and Amazon has slipped 1%, compared with a roughly 1.7% gain for the S & P 500 . Fully agentic shopping remains a longer-term prospect, and each company has characteristics that could help insulate its business as the technology develops. But those defenses look very different, providing a good cross-section of the retail space and how the rise of AI shopping brings both challenges and benefits. The biggest risk isn’t losing the sale Amazon would seem like an obvious target for disruption. If consumers can ask an external AI agent to find the best laptop, paper towels, or dog food regardless of the retailer, they may no longer need to start their search on Amazon. However, Amazon has defenses that smaller online retailers don’t — enormous scale, broad selection, fast fulfillment, and a Prime ecosystem that gives consumers reasons to keep coming back. „Amazon and [other] companies that have that strong value proposition are less exposed,“ Rosenblatt analyst Scott Devitt told CNBC. Goldman reached a similar conclusion, arguing that „platforms offering compelling prices, broad inventory selection, fast fulfillment, trusted consumer relationships, and rich first-party data assets are likely to be best positioned.“ Amazon’s massive scale has given it the flexibility to keep shoppers within its own ecosystem rather than open its catalog to outside AI agents. Notably, Amazon has blocked Muse from purchasing products directly on its platform, even as competitors like Walmart , Shopify and Best Buy have embraced the agent. The e-commerce giant isn’t sitting out the trend, either. Amazon has its own agentic shopping assistant, Alexa for Shopping , which offers personalized recommendations, product comparisons, and price histories, and can automate purchases through features such as Price Alerts and Auto-Buy. More than 350 million customers used it over the past year, while active users nearly doubled and interactions increased more than fivefold year over year in the second quarter, CEO Andy Jassy said. Customers who use Alexa for Shopping also spend more than 40% more per order on average. That’s why the concern isn’t whether AI agents will take away Amazon’s e-commerce business. The bigger question is whether Amazon can remain the place where those purchases happen while preserving the customer data and advertising economics that make each transaction more valuable. Telsey Advisory Group analyst Joe Feldman said an outside agent could create distance between Amazon and its shoppers, even if Amazon ultimately fulfills the order. In that scenario, Amazon could still get the sale while surrendering some of the valuable customer data it collects when shoppers interact directly with its platform. Advertising presents another potential pressure point. In the last few years, Amazon has leaned into its advertising business as a lucrative source of profits, with advertising revenue expected to reach $83 billion this year, according to FactSet. But that business depends on shoppers searching and browsing its platform, where brands pay to promote their products. If outside AI agents increasingly determine which products consumers see and buy, Amazon could lose some of that valuable advertising opportunity. Devitt expects Amazon could find other ways to monetize agent-driven purchases, whether through advertising within agents, higher seller fees, changes to Prime pricing, or another mechanism. But he said advertising revenue growth and sales and marketing spending are two important metrics to watch if more traffic starts arriving indirectly through agents. AI could make the value proposition more obvious Costco has built its business around membership loyalty, low prices, and getting customers into its stores, where shoppers often leave with far more than they planned to buy. An AI agent programmed to efficiently purchase items on a predetermined list could threaten some impulse spending by removing the browsing experience from the equation. There is, however, a potential advantage: an AI agent comparing prices could independently validate Costco’s value. The warehouse club is already seeing positive signs. CFO Gary Millerchip said on the company’s latest earnings call that sales originating from searches on large language models such as Gemini, Anthropic , and OpenAI are growing at a triple-digit rate, albeit from a very small base. He said the searches also highlight the value of Costco memberships and drive membership activity and engagement. That is particularly interesting because Costco does little paid digital advertising. Rather than paying to tell consumers that its products offer good value, an AI tool comparing products and prices could do some of that work for the company. „AI presents an opportunity in a sort of neutral environment for us to be able to ensure that our value and our quality shows up transparently for our members,“ Millerchip said. Costco’s curated assortment could help as well. Its buyers personally vet the relatively limited selection of items the company sells, which management said should allow its combination of price, quality, and customer reviews to stand out when consumers search using AI. There are risks. Agents could make consumers even more price-sensitive and make it easier to evaluate whether a membership is worth keeping. That makes Costco’s ability to maintain loyalty particularly important. The company’s fiscal 2026 fourth-quarter results were encouraging, with membership renewal rates improving and higher-priced executive memberships reaching an all-time high. Executive members also tend to renew at higher rates than basic members. Costco is also gaining traction with younger shoppers, an important source of future loyalty and lifetime value. The wrinkle is that younger customers are more likely to sign up online, where memberships tend to churn at higher rates than those started in stores. That makes retaining these newer customers worth watching as AI gives shoppers even more tools to continually evaluate where they get the best value. For now, however, Costco’s early experience suggests AI could become another way to demonstrate why customers pay to shop there in the first place. You can’t ask an agent to find what it can’t see The constantly changing assortment at TJX’s brands T.J. Maxx and Marshalls encourages shoppers to visit stores without knowing exactly what they’ll find — a “ treasure hunt “ that’s difficult for an AI agent to replicate. „You really can’t find an agent to find treasure,“ Wells Fargo analyst Ike Boruchow told CNBC. „You’re like, ‚I need some clothes, and I’m going to go into TJX and see if I find anything that’s fun.'“ Guggenheim analyst Simeon Siegel noted that consumers are likely to outsource „needs“ to AI before they outsource „wants.“ An agent replenishing an everyday household item is relatively straightforward, but choosing a shirt or dress is more subjective. „There is something special. There is something personal. There is subjectivity around buying discretionary items, around picking out a fall sweater, around choosing your sneakers,“ Siegel told CNBC. „Whereas there are certain areas like canceling subscriptions that having an agent work for you seems so obvious, I think it will be a while before it takes over the full process of buying apparel.“ TJX has another layer of protection: much of what it sells is difficult for an AI agent to find in the first place. Boruchow said that part of the off-price retailer’s value to brands is its ability to quietly sell excess merchandise without widely advertising those discounts online. Siegel sees TJX’s relatively small e-commerce presence as a feature rather than a flaw. „I think off-price wins because they don’t have e-commerce, not in spite of it,“ Siegel said. „How does an agent help you find something that is not searchable in the first place, how does an agent help you shop somewhere whose intention is to sell invisibly through a store?“ Like with Costco, greater price transparency could even reinforce TJX’s value proposition. If a shopper standing in a TJ Maxx can quickly determine that the item in front of them is cheaper than what they can find online, technology hasn’t eliminated the „treasure hunt“ — it has validated the treasure. For that reason, Siegel sees AI as more likely to complement the shopping experience than replace it. „I think it’ll be a tool to shop, not the exclusive way to shop,“ he said. To continue to capitalize on the in-store experience, TJX must get its product mix right. Last quarter , the company said it did not, leading to softness in its biggest division, Marmaxx, which houses T.J. Maxx and Marshalls. CEO Ernie Herrman said at the time a fix was in place, and it was bearing fruit. (Jim Cramer’s Charitable Trust is long AMZN, COST, META, TJX. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
AI agents could soon shop for consumers. What it means for retail stocks
