Will AI Recommend Your Store When a Shopper Asks?

For twenty years the web rewarded a particular kind of effort. You earned a position on the results page, made the click irresistible, and trusted a shopper to do the rest: open the tabs, weigh the options, read the reviews, and carry the deliberation to checkout themselves. The final mile belonged to the human. A shopper now opens an assistant, articulates the precise outcome they want, and watches the machine complete that mile on their behalf. It searches in real time, reasons across pages, narrows the field, and returns a finished recommendation. Often it returns a filled cart. The browse that your whole funnel was built to win never happens.

Brian Solis has been mapping that shift with unusual precision. Writing about the new wave of agentic browsers, he frames the change as a move "from a web of browsing and transactions to a web of discovery, action, and outcomes." His shorthand for what happens to the top of the journey is exact: "Discovery compresses. We move from 'find pages' to 'produce outcomes.'" The shopper expresses intent, the agent gathers sources and weighs options, and then it executes. In Solis's words, it "returns a plan, a basket, a booking, or a brief."

A basket. For a merchant, that one word is the entire story. The moment discovery resolves into an outcome, the agent has already chosen what goes in the cart, and it chose before the shopper looked at a single photograph.

The click you used to buy is becoming a completion you have to earn

Solis articulates the economic shift directly: "We're moving from pages to progress. From browsing to doing. From ad-driven clicks to agent-driven completions." A decade of growth tactics presumed you could pay to position a click in front of a person and engineer a page that converted it. When the agent does the visiting, there is no click to purchase and no page for it to land on. There is only the question of whether the system, operating on the shopper's behalf, judged your product worthy of a place in the answer.

That judgment depends on inputs most stores never prepared. When I audited the catalogs of 2,483 leading US Shopify merchants, the storefronts were polished and the photography was sharp, and almost none of it reached a machine. What an agent could genuinely read and reason over was thin, sometimes a product title and a single sentence. The shopper relocated to the assistant. The catalog stayed behind, outfitted for a visitor who was no longer arriving.

What an agent privileges when it builds the answer

Solis is specific about what earns a place in the agent's output, and the qualities he enumerates scarcely resemble the conventional playbook. "Trust becomes a ranking signal," he writes. "Agents will privilege content with provenance, recency, and clear policies over thin pages and click-bait." A reasoning system rewards the unglamorous attributes: the provenance of a claim, its currency, the clarity of your stated policies. The second requirement is harder still. "Design tilts toward machine readability," Solis writes. "If your information can't be parsed, reasoned over, and acted on by an agent, it risks being sidelined."

Sidelined is the precise fate of a beautiful page with nothing underneath it the machine can use. The agent is indifferent to your layout. It is attempting to determine whether your jacket genuinely suits a rainy commute under $150, and it can only reason from what your catalog makes legible.

A product feed confirms an item exists and shows it in stock, and then it goes quiet. It says nothing about who the piece suits, the occasions it covers, or the moment it becomes the wrong recommendation. That authored substance is exactly what an agent needs to reason its way to a confident recommendation, and it is precisely what most listings withhold.

The catalog is both the briefing and the checkout

Solis closes his essay with a to-do list for companies, and it reads almost like a catalog specification drafted in advance. Make your data callable, he advises, exposing inventory, pricing, availability, and policies through clean, well-documented interfaces. Make the underlying knowledge machine-readable through structured schema and product information an agent can genuinely ingest and reason over. Build the action surfaces as well, the carts and quotes and checkout endpoints, so an agent can complete the purchase rather than summarize a page and stop. Then maintain a current, honest account of every product. The machine has to be able to trust it.

He has a name for the discipline replacing yesterday's search optimization. "Classic SEO was written for crawlers and clicks," Solis writes, and the next move in his framing is to "design your content and commerce so agents can understand, trust, and transact." For a store, that obligation lands on the catalog, and the catalog now performs two distinct jobs simultaneously: the briefing an agent consults to decide, and the surface it operates to buy. A worked example shows the shape of that on a single page. And the editorial craft underneath it, the discipline of writing for a reader that reasons through a listing, turns out to reward demonstrable substance over the persuasion cues that twenty years of conversion copywriting trained merchants to rely on.

Whoever authored the catalog wins the outcome

The stakes concentrate on that first machine read. Shoppers arriving through an AI recommendation already convert above baseline and spend more per order, so the listing that earns or loses the agent's basket is wired directly to the most valuable demand a store now registers. The outcome the shopper receives was authored upstream, assembled by a system from whatever your catalog made legible at the instant it went looking.

Solis envisions the web rebuilding itself around agents that convert intent into completed work. The part he leaves for the rest of us to finish is the supply side of that exchange. A store does not get to decide whether the agent shops on its customers' behalf. It only gets to decide what the agent finds when it arrives: a catalog composed deliberately for a reader that reasons and acts, or one left to guesswork. Your shoppers are handing the choice to an assistant. Whether it recommends your store comes down to what your catalog gave it to read.

Sumit Jagdale is the founder of Sartorial.