You Won’t Be Able to Offload Your Holiday Shopping to AI Agents Anytime Soon
The promise of artificial intelligence revolutionizing e-commerce has captivated both consumers and industry giants, yet the vision of fully autonomous AI shopping agents remains a distant reality, especially for the upcoming holiday season. While nascent features like OpenAI’s ChatGPT allowing instant checkout on platforms like Etsy hint at a future where AI handles browsing and purchasing, these "agentic shopping" tools are still in their infancy, far from becoming the full-time virtual buyers many anticipate. Consumers hoping to delegate their entire gift list to a digital assistant this year will likely find themselves disappointed, as the technology grapples with significant hurdles and ongoing negotiations between tech behemoths and retail partners.
The current landscape of AI-powered shopping tools is characterized by a blend of ambition and practical limitations. Features currently available demand substantial user input, often operate slowly, or are restricted to a limited selection of items. This stark contrast between the futuristic vision and present-day capabilities underscores the complex challenges that developers at OpenAI, Google, Amazon, and others face. According to executives from seven tech and e-commerce companies who spoke with WIRED, the core issues revolve around mitigating costly mistakes by AI agents and establishing mutually agreeable terms for the exchange of critical product data and chat histories. Talia Goldberg, a partner at venture capital firm Bessemer, an investor in AI companies like Perplexity and Fal, candidly admits, “I haven’t yet felt a super magical agentic experience in commerce. There are big questions that have to be solved around a true functional experience.”
Despite these current shortcomings, consumer enthusiasm for AI in shopping is remarkably high. Recent surveys conducted in the US reveal a strong appetite for AI assistance: 60 percent of consumers plan to utilize AI for shopping support, while a significant 20 percent express willingness to entrust AI agents with their everyday purchases entirely. Only a quarter of shoppers, 25 percent, prefer to shop without any AI involvement. These figures, coupled with long-term projections, paint a rosy picture for the future of agentic commerce. McKinsey, for instance, estimates that agentic shopping could generate up to $1 trillion in sales in the US alone by 2030, highlighting the immense potential that industry leaders are striving to unlock.
To propel this future into motion, strategic partnerships are being forged across the industry. OpenAI has teamed up with Walmart, aiming to integrate Walmart product purchasing directly within the ChatGPT interface. Similarly, OpenAI and search startup Perplexity have secured deals with PayPal and Shopify, the latter hosting countless online stores for various brands. Google has also entered the fray, introducing AI agents capable of automatically filling out online checkout forms and even making calls to stores to inquire about pricing, demonstrating a more targeted approach to automating specific shopping tasks.
Some early prototypes, while not fully agentic, show promising signs of what’s to come. Expedia’s application for ChatGPT, for example, provides real-time flight and hotel pricing data in response to user queries. While users still need to manually complete bookings—a clear indicator that full AI agents aren’t yet involved—the feature has exceeded Expedia’s sales expectations. Clayton Nelson, a vice president overseeing Expedia’s strategic alliances with AI giants, notes, “That means there’s something in these tools that works.” This success suggests that even partial AI assistance, particularly in information retrieval and recommendation, can significantly enhance the shopping journey and drive conversions.
However, the path to widespread AI adoption in commerce is fraught with challenges, some of which echo past difficulties in digital retail. Social commerce, or shopping directly through platforms like TikTok and Instagram, has struggled to gain significant traction in the US, partly due to persistent consumer distrust of tech giants and the reluctance of large retailers to fully embrace these platforms. To prevent AI initiatives from encountering similar resistance, major payment processors such as Visa and software startups like New Generation, which specialize in developing chatbot solutions for stores, are actively working to broker technical compromises with retail partners. Adam Behrens, CEO of New Generation, believes that service providers like his company can build trust with retailers more effectively, a crucial element for successful collaboration.
Retailers are keen to participate in this evolving ecosystem because chatbots have rapidly become indispensable tools for consumers researching and validating purchases. By partnering with AI companies, retailers hope to ensure that chatbots not only deliver accurate product information but also consume fewer computing resources when executing online orders. Such efficiencies could translate into boosted profits for both AI developers and retailers, provided they can successfully navigate the intricate negotiations and technical integrations.
The current state of agentic shopping has not escaped criticism from top industry figures. Amazon CEO Andy Jassy, in a recent earnings call, voiced his dissatisfaction with how agentic shopping currently functions on other platforms. He critiqued the lack of personalization, the absence of integrated shopping history, and frequently inaccurate delivery estimates and prices. Jassy emphasized the need to "find a way to make the customer experience better and have the right exchange of value." These comments highlight the significant gap between existing AI shopping experiences and the seamless, personalized interactions consumers expect from leading e-commerce platforms. A practical demonstration of these limitations was observed in a WIRED test, where Opera browser’s AI agent took a ponderous 45 seconds to add eggs to an Amazon cart—a task that a human could complete manually on Amazon’s shopping app in just one-third of that time.
Companies like Opera are proactively engaging potential partners through workshops to collaboratively address security and design choices. Per Wetterdal, an executive vice president leading Opera’s commercial partnerships, stresses the importance of interoperability: “If our agent doesn’t work with the biggest websites people go to, it will be a suboptimal experience. No one benefits if [a purchase] is ending up at the wrong place or in the wrong quantity.” This collaborative approach underscores the industry’s recognition that widespread adoption requires robust, reliable, and secure integrations across diverse retail environments.
At the heart of these ongoing negotiations are two critical elements: money and data. The financial aspect of agentic shopping could be relatively straightforward, with AI companies like Opera seeking a percentage of sales for facilitating purchases. OpenAI, for instance, has already established a precedent by collecting a “small fee” from partners like Etsy for Instant Checkout transactions. This commission-based model offers a clear path for AI companies to monetize their services.
However, data sharing presents a far more complex challenge. Retailers fiercely guard their pricing, availability, and customer data, recognizing them as proprietary assets crucial for maintaining a competitive edge. Conversely, AI companies aim to protect conversation histories to preserve the sense of intimacy and personalization that chatbots can offer. Yet, for chatbots to effectively fulfill user requests, they require real-time product information, while retail brands desire richer context to cultivate deeper relationships with shoppers. OpenAI’s app feature, for example, provides partners like Expedia with a user’s IP address and relevant chat queries, subject to user permission. Expedia’s Nelson, while satisfied with the initial data exchange, ultimately desires more comprehensive information, such as whether guests are friends, their travel history, or other preferences, provided users consent. This desire for deeper insights highlights the tension between data privacy and the quest for hyper-personalized shopping experiences.
The competitive landscape is also heating up, occasionally leading to legal battles. Amazon recently sued Perplexity, alleging that its AI agents interfered with Amazon’s businesses, including advertising sales and Prime subscriptions, by making purchases on users’ behalf. Perplexity has vowed to fight the lawsuit, underscoring the legal ambiguities and potential conflicts as AI agents navigate the existing e-commerce infrastructure. Simultaneously, Amazon is developing its own shopping agent, codenamed “Buy for Me.” This feature leverages agentic AI to complete purchases on other retailers’ websites when an item is out of stock on Amazon. While the tool automatically adds items to a cart, allowing users to check out with their Amazon payment details, it notably prevents third-party stores from receiving shoppers’ real email addresses, and stores can even contact Amazon to block its agents. This demonstrates Amazon’s strategic intent to control the customer experience and data flow, even when facilitating purchases off its primary platform.
Despite the hype and significant investments, many in the retail sector believe that fully functional agentic shopping solutions are still nascent. An executive for a large clothing retailer in California, who preferred to remain anonymous, expressed eagerness for such deals due to the significant traffic chatbots are driving, yet lamented the underdeveloped nature of current offerings. “Up to today, no one has a solid solution,” the executive stated, suggesting that many announcements are more about marketing than mature technology.
Some smaller AI companies are deliberately holding back from pursuing full agentic shopping deals. Archit Karandikar, CEO of CoInvent AI, which develops the travel planning chatbot Airial, emphasizes that generating useful recommendations with AI is a substantial challenge in itself. Adding agentic purchases, given the current state of technology, would be too much too soon. “You can’t be spending someone’s money without being sure you’re making the right transactions,” Karandikar asserts. Consequently, Airial currently links to booking websites and earns a commission when a user makes a purchase, maintaining a human-in-the-loop approach.
Expedia’s Clayton Nelson echoes this sentiment with a blunt assessment of the risks involved. “My goodness, no one wants to mess up their vacation for their entire family because a bot went left instead of right, or didn’t follow the specific prompt that was given,” he warns. “It’s up to us and our partners to make sure that we never leave travelers astray. And so that’s the big thing that’s holding us back on fully agentic booking experiences.” The implications of an AI making a costly or incorrect purchase, especially for high-stakes items like travel or significant holiday gifts, are too great to overlook.
This holiday season, while chatbots will undoubtedly assist users in selecting gifts, adding them to carts, and perhaps even facilitating instant checkouts, humans will remain firmly in control of the final decisions. The vision of a fully autonomous AI agent handling all holiday shopping chores, from recommendation to purchase and delivery coordination, is still a few years away. Perhaps next year, consumers can finally blame the bots if a loved one doesn’t like their gift. For now, the intricate dance between human oversight and AI assistance continues, slowly but surely paving the way for a more agentic future.
