If AI shopping agents start cheap, why would stores ever agree to pay them more later?

This explores why sellers might agree to give AI shopping agents a bigger cut of each sale over time, even though the agent starts out taking almost nothing. The clearest picture of that path is Jaipuria's three-stage model. An agent begins by earning payment commissions of under half a percent. It then adds affiliate fees of around 4%. In categories where it fits, it ends up taking a 12% cut the way an aggregator does. The blended cut climbs from roughly 0.8% toward 4% How could an AI agent's revenue per transaction grow over time?. The model shows how the cut grows. It says less about why merchants go along with it, so the corpus gives a partial answer that has to be pieced together from nearby material.

Travel has the strongest real-world example. Many expected AI agents to cut out online travel agencies like Expedia. Instead, independent hotels' dependence on those agencies rose to 63.4%. The agencies stayed in the deal as the merchant of record, the party that legally sells to the guest, by plugging into AI agents through APIs, and Expedia's business-to-business segment grew 24% Are AI agents actually removing online travel agencies from bookings?. Merchants accept high fees when the middleman controls access to customers they can't easily reach themselves. Once agents become the place where purchases start, the same pull applies. Consumer trust points the same way: people trust a known intermediary to handle agent payments far more than they trust the AI on its own Does naming a payments brand increase trust in AI transactions?. Sales flow toward whoever shoppers trust, and merchants follow the sales.

A second reason is that the money may come out of a different budget. Jaipuria argues that when agents rather than people do the shopping, ads lose much of their value, because an agent compares value instead of looking at page position. Merchants would then move ad spending into lower prices and better terms Could agents redirect merchant ad budgets toward customer savings?. Seen this way, a higher take rate isn't a new cost. It is the old customer-acquisition budget showing up in a new place. That has a less comfortable side. Jaipuria also warns that merchants' payments for placement could be folded into take rates where users can't see them. That would quietly give the agent a reason to recommend whoever pays most Could agent advertising hide inside platform take rates?.

This is where the question matters for shoppers. Rising take rates paid by merchants are the mechanism that Azhar and the MIT 'Coasean Singularity' authors flag: an agent funded by merchant fees cannot be fully loyal to the buyer, and whoever builds and pays for the agent sets its goals Can an AI agent serve both merchant and user interests fairly? Does agent ownership determine whose interests get optimized?. Anthropic's Project Deal suggests users may not notice. People represented by weaker agents got measurably worse prices but rated their deals as just as fair Do people notice when their agents secure worse deals?. A slow shift toward serving sellers could go unnoticed. That is part of why Owens and Law argue for fiduciary duties now, before these fee structures settle in Should AI shopping agents face fiduciary duties like human brokers?.

One counterweight is worth knowing about. Merchants are also being asked to carry more risk. Agent purchases remove the behavioral signals merchants used to catch fraud, like typing rhythm, yet they stay liable as merchant of record Does agentic commerce shift fraud liability onto merchants?. Merchants paying more and also carrying more risk suggests agent platforms are gaining real bargaining power. Merchants would accept higher take rates less because the deal is good and more because, as with travel agencies, there's no other way to reach those customers.