For a small but fast-growing share of purchases, the shopper never opens the product page at all. An AI agent does that instead, comparing competing listings, checking the return policy in a fraction of a second, and completing the purchase on a card the customer pre-approved days earlier. It already happens through tools like ChatGPT’s shopping features, just not yet at scale. Today, only 3 percent of purchases in the US and UK currently move through an AI agent, though a third of consumers expect at least one in ten of their own purchases to be AI-driven within the next year. The gap between those two numbers is where brand loyalty is now being decided.
The Death of the Browsing Session
For two decades, brand marketing has leaned on the browsing session, a person scrolling, comparing open tabs, and forming an impression from photography, copy, and site speed before deciding what to buy. AI shopping agents skip most of that process. They read structured product data instead of homepage design, and they weigh price, shipping time, and return terms instead of a hero banner.
Pablo Gerboles Parrilla, who has spent recent years building performance marketing systems for clients through Pabs Marketing, argues that many brands are still marketing to the shopper of five years ago while the actual buyer has become software that never sees a single ad. That shift changes what a marketing budget is supposed to accomplish, because persuasion has far less to work with when the decision-maker can’t be charmed by a photograph.
Brand Loyalty Is Already Negotiable
Checkout.com’s research found that 57 percent of consumers would let an AI agent switch them to a different brand if it found better value, a figure that should concern any marketing team that has spent years building preference through loyalty programs and repeat-purchase discounts.
Trust has not caught up with that willingness. Twenty-seven percent of consumers say they trust no organization to operate an AI shopping agent on their behalf, and 24 percent say they will never delegate a purchase to one at all, according to a mid-2026 payments industry survey.
A meaningful share of buyers are ready to hand over the decision the moment they trust the agent making it, and most brands have no visibility into what tips that trust one way or the other.
Agent-Readable Catalogs Are the New Storefront
Retailers are already adjusting to a buyer that reads code instead of copy. OpenAI’s Instant Checkout, built on its Agentic Commerce Protocol, now lets ChatGPT complete purchases inside a conversation for sellers on Etsy and a growing list of Shopify merchants, including Glossier, Spanx, and Vuori.
The mechanism reduces a purchase to structured product data, a price, a return policy, and a fulfillment promise, then lets the agent decide. A brand’s visual identity, the element most marketing budgets are built around, plays no role in that decision. What matters is whether the catalog feeding the agent is complete, accurate, and current, and whether it stays that way after the marketing team stops paying attention to it.
What Pabs Marketing Tells Clients to Fix First
Gerboles Parrilla says the mistake he sees most often is treating this shift as an IT job when it belongs to marketing.
“We’re not just building companies, we’re building long-term, resilient ecosystems around powerful ideas,” he said, describing the same logic he now applies to client catalogs feeding AI agents.
In practice, that means auditing product feeds the way a team once audited a homepage, checking that return windows, shipping estimates, and stock levels are accurate down to the hour, because an agent that catches one wrong data point will simply choose a competitor’s listing instead.
Price Is Not the Only Variable the Agent Weighs
Groceries and household goods lead the categories where consumers already say they would delegate to an agent, at 41 percent and 31 percent respectively, while financial services lag at 15 percent, a reminder that trust and category risk still shape how far this shift goes.
Even within categories where trust runs high, price is one input among several. Shipping speed, return friction, and stock accuracy all factor into what an agent recommends, which means the most exposed brands are those that optimized for a browsing customer’s emotional response rather than the operational facts an algorithm can verify.
A brand that wins on design but loses on return policy clarity may never make the shortlist an agent actually compares.
AI Rewards Judgment, Not Just Automation
“AI won’t replace good judgment. It’ll amplify it,” Gerboles Parrilla said.
An agent can execute a pricing or catalog decision instantly, but it still needs someone to decide which data points matter and why. The brands adapting fastest are not necessarily the ones racing to plug into every new checkout protocol. They are the ones treating agent readiness as an extension of the discipline that used to govern a well-run marketing funnel, built on accurate data, consistent execution, and a clear read on what actually earns the sale.
Loyalty Now Lives in the Data Layer
The marketing instinct is still to defend brand loyalty with better storytelling. The numbers coming out of 2026 point to a layer underneath that story: the catalog data, the fulfillment terms, and the trust signals that determine whether an agent recommends a brand at all.
A shopper who never scrolls past the fold still forms a loyalty of a kind, one built entirely on whether an agent’s calculation kept landing in that brand’s favor. Companies that correct their data before competitors do can turn that calculation into an advantage most marketing budgets were never built to measure.
