AI Shopping Agents Are Coming — But the Industry Isn't Ready Yet
Imagine waking up, telling your AI assistant you need a new pair of ski gloves, and having it research, compare, and purchase the best option on your behalf — all before your morning coffee is done. That future is closer than most people think. But according to industry leaders and experts who spoke at Fortune Brainstorm Tech 2026, the road between "AI can help you shop" and "AI can shop for you" is still riddled with obstacles that no one has fully solved.
AI shopping agents — autonomous software programs capable of browsing, selecting, and purchasing products on a user's behalf — are rapidly emerging as the next major frontier in e-commerce. Yet a growing consensus among technologists, entrepreneurs, and legal experts suggests the infrastructure, standards, and policies needed to support them simply aren't in place. Here's a deep dive into where things stand, what's holding the industry back, and why the stakes couldn't be higher.
What Is an AI Shopping Agent?
An AI shopping agent goes well beyond a simple product recommendation engine. While most people have used AI models like ChatGPT to search for products they might want to buy, a true shopping agent would take action on your behalf — comparing prices, verifying availability, and completing a transaction without requiring the user to click through a single checkout page.
The appeal is obvious. In an age where consumer attention is increasingly scarce, automating the tedious parts of online shopping could save both time and money. For retailers, an AI agent that reliably converts browsing into buying could represent a massive new sales channel. But as experts at Fortune Brainstorm Tech made clear, the gap between the concept and the reality is still enormous.
The Core Problem: Today's AI Is Still Falling Short on Commerce
Even at the most basic level of product discovery, existing AI models are underperforming. Melissa Bridgeford, cofounder and CEO of Wizard Commerce — a company that builds AI shopping agents — pointed out a striking statistic during the panel: when a user asks ChatGPT about a product category, such as ski gloves, the chatbot responds with specific product recommendations only 9% of the time. That's a staggeringly low hit rate for a technology that is supposed to be transforming commerce.
This means that even before we get to the more complex problem of completing a purchase, today's leading AI models are struggling to consistently perform the most fundamental task in the shopping journey — telling you what to buy. For businesses and developers building on top of these models, that gap represents both a challenge and a significant market opportunity.
OpenAI's Stumble: A Cautionary Tale for Agentic Commerce
One of the most talked-about moments in the recent history of AI-powered shopping was OpenAI's attempt to embed commerce directly into ChatGPT through an Instant Checkout feature. The feature was designed to allow users to complete purchases directly from the chat interface — a genuinely promising step toward the agentic commerce future many are envisioning.
However, Bridgeford was candid about what she saw as OpenAI's misstep: the company pivoted away from Instant Checkout before it could gain meaningful traction. The fallout was swift. Early retail partners, including Walmart, pulled out of the relationship following the pivot. The episode serves as a cautionary tale about how quickly momentum can be lost in the agentic commerce space, and how critical it is to follow through on platform commitments in order to build retailer trust.
Three Major Barriers Blocking the Rollout of Shopping Agents
Matt Maher, founder and CEO of M7 Innovations, an independent tech research and development firm, outlined the structural challenges that are preventing AI shopping agents from operating at scale today. They fall into three broad categories:
- Security protocols: Most online retailers have built security infrastructure designed to detect and block non-human interactions. An AI agent attempting to complete a purchase on a user's behalf will often run into authentication walls, CAPTCHA systems, and fraud detection algorithms that flag the behavior as suspicious or unauthorized.
- Lack of agentic commerce standards: There is currently no universal framework defining how AI shopping agents should identify themselves, communicate with retail systems, or authenticate a user's consent to make a purchase. Without agreed-upon standards, every retailer integration becomes a custom engineering problem.
- Retailer policy barriers: Many retailers have proactively moved to block third-party shopping agents from their platforms entirely. Whether driven by concerns about fraud, loss of customer data, or the desire to maintain control over the checkout experience, these policies represent a significant structural headwind for the entire industry.
Is the Industry Moving Toward Consensus?
Despite these challenges, there are reasons for cautious optimism. Bridgeford noted during the panel that she believes the broader industry is beginning to coalesce around allowing agentic commerce to proceed. The business case is simply too compelling for retailers to ignore forever — an AI agent that can reliably drive purchases on behalf of millions of users represents a potentially enormous source of incremental revenue.
That said, there remain significant unresolved questions that must be answered before widespread adoption can happen. Chief among them are issues around fraud prevention, refunds, and returns. When an AI agent makes a purchase that a human customer later disputes — whether because the agent misunderstood the user's preferences, selected the wrong size, or was itself compromised — who bears responsibility? How does a retailer process a return initiated not by a person, but by a piece of software? These are not hypothetical edge cases; they are real operational challenges that will need robust, industry-wide solutions.
What Needs to Happen Next
For AI shopping agents to move from a promising concept to a mainstream retail reality, several things need to happen in parallel. The technology itself must improve, particularly around accurate product discovery and contextual understanding of user intent. Industry stakeholders — including AI developers, retailers, payment processors, and regulators — will need to come together around shared standards for how agentic transactions are authorized and verified. And perhaps most importantly, a trust framework will need to emerge that gives both consumers and retailers confidence that AI-driven purchases are secure, accurate, and reversible when things go wrong.
The potential upside is enormous. A well-functioning agentic commerce ecosystem could reshape the entire e-commerce landscape, shifting power from search-driven discovery models toward AI-mediated relationships between consumers and brands. But as the Fortune Brainstorm Tech panel made clear, that future won't arrive on its own. It will require deliberate, collaborative work — and the industry is only just beginning to have the right conversations.
The Bottom Line
AI shopping agents represent one of the most exciting — and complex — frontiers in modern technology. The vision is compelling: software that shops smarter, faster, and more efficiently than any human could. But the gap between vision and reality remains wide. Security barriers, missing standards, retailer resistance, and unresolved questions around fraud and returns are all standing in the way. The technology is coming. The question is whether the industry can build the infrastructure fast enough to meet it.