If outside apps can build one shopping cart from many stores, should your brand build its own AI shopping assistant first?

This explores whether brands can protect themselves by running their own shopping agents, at a time when third-party agents might fill one cart from many stores. Here's the threat. Bain argues that agents able to shop across several vendors, and not just pick products within one store, could break up the idea of the retailer-owned basket. Grocery is most exposed. Competition would then happen at the level of the whole basket, with prices fully visible and margins squeezed Could AI agents reassemble shopping baskets across multiple retailers?. Accenture's answer is to become the agent of choice so you aren't left being picked by somebody else's agent. Its survey found consumers say they prefer agents built by retailers Will brands succeed by building their own AI shopping agents?. That evidence is stated preference, though, not proof that brand-built agents win once they're actually in use.

The argument depends on one idea: whoever owns the agent decides what it optimizes for. The Coasean Singularity paper calls platform-built agents 'bowling shoe' agents. They're rented from the house, so they serve the seller by default. Agents owned by consumers can be set up around what the buyer wants Does agent ownership determine whose interests get optimized?. This cuts both ways for brands. Owning the agent puts your products in front of the shopper. But a shopper who knows the agent is the house's agent has every reason to bring their own instead. Hedges points out the same conflict in ad-funded platform agents, which have built-in reasons to favor paid placements Do advertising-funded shopping agents favor paid product placements?. Owens and Law, and Senator Warner's draft bill, would go further and impose best-interest or fiduciary duties on agents. That could make a brand-tilted agent legally risky, not just less trusted Should AI shopping agents face fiduciary duties like human brokers? Could federal AI agent registries solve platform access unfairness?.

So the corpus points to protections other than owning the agent. One is brand preference the shopper states up front. In Accenture's survey, 56% of consumers would tell their agent which brands to buy, and 37% of loyal customers wouldn't switch Do consumers actually want to delegate shopping decisions to AI?. A brand that's written into the shopper's instructions survives any agent, whoever owns it. Another is trust borrowed from a familiar name. Only 23% of US adults trust AI to make payments, but 61% trusted Visa to handle payments that agents start Does naming a payments brand increase trust in AI transactions?. A known name works like a reassurance layer, and that may matter more than owning the agent.

The less comfortable finding is that the fight may move from owning the agent to persuading it. In the ACES experiments, a seller-side agent that only rewrote product descriptions raised its product's market share by 3.7 to 14.9 percentage points with most of the buyer models tested. Price, ratings and position stayed the same Can AI seller agents shift buyer preferences through description edits?. Jaipuria argues merchants will move ad budgets into real price cuts and better terms, because agents look at value, not page position Could agents redirect merchant ad budgets toward customer savings?. In a world of mixed-vendor baskets, then, brands may be protected less by owning the shopper's agent and more by being legible and competitive to whatever agent shows up. BCG's estimate is a sobering backdrop: in its view, consumers and networks capture most of the gains while merchants face built-in disadvantages Who wins and loses as AI agents reshape shopping?.

One thing the brand can't control may decide how much any of this matters. Waller expects agentic commerce to split into open and closed systems, much as payment networks did Will agentic commerce platforms adopt open or closed models?. In a closed ecosystem, a brand's own agent could be a real moat. In an open one, where any agent can shop anywhere, it's just one option among many, and the shopper's agent may simply route around it.