Chewy's ~$50M AI Bet and Its Clear Answer to Agentic Commerce: How 84% Autoship Defends Against Disintermediation

Chewy's Clear Answer to Disintermediation

Chewy Has a $50 Million AI Bet and a Clear Answer to Agentic Commerce

If you're selling a commodity, I think the disintermediation issue is likely one that needs paying attention.

aimmediahouse.com

AI agents that search for, compare, and buy products on a consumer's behalf. As this "agentic commerce" becomes real, the thing retailers fear most is disintermediation. An AI agent routes a consumer who would have visited a retailer's own site to a brand's DTC store or a competing platform instead. In a world where being surfaced in results determines sales, that is a genuine threat.

Chewy, the US pet e-commerce leader, has offered an unusually specific answer to this question. On its fiscal Q4 2025 earnings call (reported in March 2026), CEO Sumit Singh framed agentic commerce not as a threat but as a "future incremental demand and distribution channel." While most retailers scramble to be surfaced by AI agents, Chewy's argument points the other way: its best customers do not need to be surfaced at all.

The basis for that claim lies in the structure of the business itself. On the disintermediation debate, Singh put it this way.

"

If you're selling a commodity, I think the disintermediation issue is likely one that needs paying attention. But from that point of view, we believe Chewy is quite well insulated, given our value proposition is not primarily search aggregation and because our customer relationship is not primarily built around onetime discovery.

Source: Sumit Singh, Chewy CEO

"Search aggregation" here refers to a business that works by getting picked from among the products lined up in search results. That is precisely the type of business most destabilized when an AI agent steps in between. Chewy's logic comes down to a single point: it does not depend on that.

The Line of Defense Is Autoship—84% Have Graduated From Discovery

The number behind Chewy's argument is its subscription rate. 84% of the company's net sales run through Autoship, its automated recurring delivery service. For full-year 2025, the figure was 83.3%, or roughly $10.5 billion in sales from Autoship (Digital Commerce 360). A customer with their dog food, flea and tick medication, and prescriptions on recurring delivery no longer asks an AI agent, "Where should I buy this?" They have already decided.

Why does this structure become a wall against disintermediation? Agentic commerce bites at the top of the funnel—the discovery and purchase-decision stage. To borrow the original article's framing, Chewy's most valuable customers have already exited that stage. Their relationship with Chewy is not a "discovery relationship" where they search for goods all over again each time, but a "service relationship" where set items arrive on a set cadence.

Repeat behavior is already locked in. That single fact structurally neutralizes the intervention of search or AI agents. A one-time transaction leaves room to slip in an alternative from the side, but a habit baked into recurring delivery gives an agent almost no opening. Agentic commerce is an "increment" rather than a "threat" for Chewy precisely because its existing recurring revenue is hard to erode.

That said, this does not mean Chewy is ignoring the industry's tectonic shift. The company has stated plainly that it is "closely following" agentic commerce internally and is considering implementing it on its own platform. Pet is an emotional category where trust, relationships, and empathy matter. Combine that with leadership in price and selection, plus ongoing convenience in both purchase and delivery, and the competitive position holds—that is Singh's read.

The ~$50 Million AI Bet Goes Toward Efficiency

If the stance on agentic commerce is the strategic story, the financial story lies in the substance of the AI investment. What Singh disclosed was a target of more than $50 million in annualized efficiency savings in 2027. For 2026, the company first expects "low tens of millions" of dollars in efficiency gains, with a meaningful step-up in 2027.

What stands out is that this investment is aimed not at flashy customer-facing features but at the efficiency of internal operations. Six deployments are already said to be running in production, spanning customer service, fulfillment, pharmacy, marketing operations, campaign optimization, and creative optimization. The thinking is to use AI not only as a tool to grow sales but as a tool to structurally lower the cost to serve.

The most concrete result cited was a self-service refund-and-returns tool. Launched eight weeks before the March earnings call, it lets customers initiate refunds and returns without contacting a human agent. It is designed to reduce handle times, contact rates, and cost to serve simultaneously. Singh called the tool's engagement and success rates "quite impressive"—measured language for a CEO who tends toward specificity, but a sign of traction nonetheless.

The $50 million figure comes from several effects compounding. AI helps agents extract information and deliver consistent service with less effort. As a result, agent experience improves, turnover falls, and cost to serve declines—all at once. The scale of the efficiency emerges where those three compound together. Chewy's long-range plan targets a 10% adjusted EBITDA margin, and AI-driven SG&A efficiencies are positioned as part of the path to close the remaining gap of roughly 350 basis points.

The Evidence That "AI Works" Is Piling Up

Behind Chewy's confidence is an industry where agentic commerce is shifting from "theory" to "evidence." In May 2026, a Fast Company investigation quoted Google and OpenAI's commerce leads saying the tipping point where AI shopping becomes commonplace is months away, not years.

Real numbers are emerging too. On its Q1 earnings call, Walmart disclosed that users of its AI shopping assistant Sparky spend about 35% more per order. Lowe's AI assistant Mylow has reportedly converted at several times the rate of non-users. On the payments side, Mastercard launched Agent Suite for merchants, and Visa stood up agentic commerce infrastructure betting its entire network on the shift. Week by week, the agentic commerce conversation is gaining the backing of concrete cases.

In this environment, Chewy articulated its own position early. Where many retailers have yet to sort out whether they sit on the disintermediated side or not, Chewy stands out for answering the question with a structurally specific argument.

Implications for E-Commerce Operators

What Chewy's case shows is that preparing for agentic commerce is not only about "optimizing to be surfaced by AI agents." The real question is how much you have built up customer relationships that resist agent intervention.

The more a business sells commodities through one-time search, the higher its disintermediation risk. Conversely, operators with subscriptions, expertise, and ongoing service relationships gain a structural moat for the age of AI agents. How much of your sales come from "customers who have graduated from the discovery stage"? That ratio is one yardstick for measuring resistance to agentic commerce.

At the same time, Chewy is not playing pure defense. It directs most of its AI investment toward lowering cost to serve and recycles the savings into competitive price and quality. It takes agentic commerce on board as a future incremental channel while shielding existing recurring revenue from agent intervention. This two-pronged stance is the substance of Chewy's "clear answer."

Conclusion

What makes Chewy's earnings interesting is the articulation of its AI strategy more than the financial figures. Against a backdrop of a ~$50 million AI efficiency target and an 84% subscription structure, the company positioned agentic commerce as an incremental opportunity rather than a threat. What sets it apart from other retailers is the concrete answer it gave to the disintermediation debate: its best customers have already graduated from the discovery stage.

For e-commerce operators thinking about agentic commerce, Chewy's case poses a question. Is your relationship with customers a "one-time discovery" that agents can interrupt, or an "ongoing service" that they cannot? That distinction will shape competitiveness over the next several years. From here, attention turns to how Chewy implements agentic commerce on its own platform and how far it actually delivers on its AI efficiency target.