When AI Does the Shopping: Agentic Commerce and the Redesign of Marketing

Search for a product, compare, choose, buy — into this long-familiar sequence, AI has begun to insert itself. In 2026 the shopper's entry point is quietly shifting from "search it yourself" to "let AI handle it." This is not only a retail or e-commerce story. How you get found in search, how you run ads, how you structure product information — the very foundations of marketing are being forced into a rebuild. This piece maps the reality of what's called agentic commerce and the preparations a business should make.

From "searching" to "delegating"

According to the retail platform Mirakl, more than 70% of shoppers have already folded large language models (LLMs) into their buying — for gift inspiration, price benchmarking and product analysis. Beyond that, retailers such as Amazon (with "Rufus") and Lowe's (with "Mylow") have begun placing their own AI assistants at the entrance to product discovery. A shopper states conditions in chat, and the AI presents candidates. That recommendation slot becomes the new place for advertising.

At that point a brand's contest shifts from "what rank do I appear at in search results" to "am I chosen in the AI's answer." If you are not recognized by the AI before the shopper ever sees you, you don't even make the field.

Search rank and AI citation became different things

The change is crystallized by Google's May 2026 core update. By the analysis, the information AI cites in its answers is decoupling from traditional search rank. AI pulls not only from top pages but from deeper search results, forums, industry experts and structured content. Ranking high and being cited by AI are no longer the same.

Content that gets cited shares common traits: a question-answer format, tight topical focus, extractable passages, consistency with trusted sources, and demonstrated expertise. A design that meets these is less an extension of traditional SEO than a new craft of optimizing for AI — so-called GEO (generative engine optimization).

The foundation marketers must rebuild — product data

As the precondition for being chosen by AI, Mirakl repeatedly stresses getting product data in order. For AI to recommend accurately, it needs "structured, comprehensive and accurate product attributes." If size, material, use, audience and price stay vague, you won't make the AI's shortlist. A human shopper fills small gaps with imagination; an AI may treat a missing attribute as simply "no match." Before dressing up a product page for people, get it into a form machines can read correctly — that is the first step of the new shelf-space game.

Measurement and retail media become the main battleground

The ad world is shifting too. Retail media — the ad inventory retailers own — is projected to exceed $176 billion by 2028. Yet only 12% of operators in North America and Europe have reached genuine full-funnel operation across on-site, off-site and in-store (Koddi/Forrester, as of November 2025). The upside is large, but most firms have not yet mastered it.

Here everyone agrees on the importance of measurement. 86% of decision-makers in North America and Europe call strengthening measurement and attribution to prove ROI a high or critical priority (same survey). AI has begun to be used to raise that measurement's precision: Sam's Club, for instance, uses an AI measurement tool to surface a 12-month multichannel performance view grounded in membership data.

Ad operations move to "converse and delegate"

The shape of operation itself changes. Google launched "Ask Advisor," built on Gemini, connecting Google Ads, Analytics, Merchant Center and more into a single conversational interface, so advertisers get cross-system answers by asking rather than hopping between tools. OpenAI, too, opened ChatGPT advertising to all U.S. businesses. The advertiser hands over assets, goals and a budget, and the AI takes on targeting, bidding and creative optimization — that mode is becoming the norm. The practitioner's job shifts from tweaking settings to designing what to delegate to the AI and what to decide.

What's next — AI stands between customer and brand

What emerges across all of this is a structure in which a new intermediary, the AI, stands between customer and brand. The shopper asks the AI, the AI chooses, and ads compete inside the AI's recommendation slot. Fail to be recognized correctly by this layer, and even an excellent product may as well not exist. That is exactly why the unglamorous foundation work — getting product data in order, designing information to be cited by AI, sharpening measurement — pays off more than ever. In an age where inflow is decided less by "what search rank" than by "can AI find you," is your shelf visible to the AI, correctly?

Key takeaways

In 2026 the entry to shopping is shifting from "search it yourself" to "let AI handle it." By Mirakl, over 70% of shoppers have already folded LLMs into buying, and retailers' own AIs like Amazon Rufus and Lowe's Mylow have become the discovery surface. Google's May core update decoupled AI citation from search rank, putting a premium on information design that answers questions directly, narrows the topic and shows expertise. Retail media will grow past $176 billion by 2028, even as only 12% reach genuine full-funnel operation (as of November 2025). As Google's Ask Advisor and ChatGPT's U.S.-wide ad opening show, operation is moving to "converse and delegate." What decides the outcome is the foundation: product data in order and information designed to be found by AI.