When the Agent Does the Shopping, Where Does Your Brand Stand?

Something shifted at the start of 2026 that the industry has not yet fully reckoned with. Within weeks of each other, Google unveiled its Universal Commerce Protocol at NRF with Walmart, Target, and Shopify as co-developers, Amazon made Alexa+ available free to all 250 million U.S. Prime members, and Walmart began formally serving sponsored ads inside Sparky, its in-app AI shopping agent. Each of these moves, taken individually, would register as a significant product milestone. Taken together, they mark a change in the basic mechanics of how consumer goods get discovered, evaluated, and purchased.

The consumer data was already ahead of the industry. As of January 2026, 41 percent of consumers had used a dedicated AI platform for product discovery. More telling: 33 percent say they have fully replaced their previous discovery methods. They are not running AI searches alongside traditional ones. The prior behavior is simply gone. Among early adopters and AI power users, that replacement rate climbs above 50 percent. For brands that built their growth model on keyword strategy, sponsored search, and digital shelf optimization calibrated to human browsing behavior, the ground has moved.

The Infrastructure Is Already Here

Industry conversations about agentic commerce have too often been framed around consumer readiness, as though adoption were the variable that would determine pace. The more pressing constraint right now is on the brand and retailer side: whether the product data, content infrastructure, and media frameworks exist to compete in agent-driven transactions.

Google’s Universal Commerce Protocol is perhaps the clearest signal of how serious and how fast this has become. Co-developed with Shopify, Walmart, Target, Wayfair, and Etsy, and endorsed by Mastercard, Visa, Stripe, Best Buy, Home Depot, and more than twenty other commercial partners, UCP is an open standard designed to allow AI agents to manage the full shopping journey from discovery through checkout within a single conversational flow. The headline feature is native checkout inside Google Search, AI Mode, and Gemini, allowing a shopper to complete a purchase without ever leaving the conversation. For retailers, the protocol also enables personalized offers, loyalty enrollment, and branded AI sales agents embedded directly in Google Search results.

Google’s Shopping Graph now indexes more than 50 billion product listings, updated at a rate of two billion per hour. That infrastructure, combined with UCP and the agentic checkout capability Google first piloted in late 2025, is a serious attempt to stop being the top-of-funnel bridge that sends shoppers elsewhere and become the destination where transactions complete. The conversion gap between Google Shopping and Amazon has historically been wide: Google’s shopping ads convert at roughly two percent, compared to Amazon’s marketplace rate of nearly ten percent. Closing that gap is the actual strategic ambition behind UCP, even if the protocol language is framed in terms of openness and interoperability.

For Walmart, the story runs on two parallel tracks. Sparky, the customer-facing AI shopping agent inside the Walmart app, is already shaping purchase decisions at real volume. Walmart’s own survey data found that 81 percent of customers who have used Sparky engaged with it to check product availability and review specifications before buying. Sponsored prompts within Sparky are now live, meaning a shopper asking a conversational product question may receive a brand-paid response woven directly into that answer. This is retail media operating in a fundamentally different register than the banner ad or keyword auction. The question is no longer about impression share on a search results page. It is about whether your brand earns a mention when Sparky is deciding what to recommend.

On the supplier side, Walmart has simultaneously launched Marty, an AI advertising assistant that helps brands build, manage, and optimize Walmart Connect campaigns through conversational prompts. Walmart reports that 97 percent of the queries Marty receives are unique, suggesting brands are using it for highly specific campaign decisions rather than templated requests. Both tools together represent Walmart’s claim on the agentic infrastructure for its ecosystem: Sparky shapes what shoppers discover, Marty shapes what brands can do about it.

What Changes When No Human Is Doing the Searching

The implications for CPG brands go deeper than a new ad format or a revised content checklist. Traditional digital shelf strategy was built around human browsing behavior: searchable titles, keyword-rich copy, image carousels designed to stop a scroll, and ratings architectures calibrated to visible social proof. An AI agent processing a shopper’s prompt does not scroll, does not respond to lifestyle photography, and does not weigh a headline the way a person skimming a search results page does. It reads structured data, evaluates attributes, and matches against the parameters of the request.

Google Cloud’s published guidance for CPG companies makes the stakes concrete: product attributes that are not structured and tagged are effectively invisible to agents. A brand whose packaging carries a sustainability certification that is not embedded as a verifiable data attribute in its product catalog will not surface when an agent is executing a query for “verified sustainable packaging.” The same applies to dietary certifications, performance claims, compatibility data, and any other attribute a consumer might express as a natural language preference. What has historically been a marketing decision handled through copy and creative now requires clean, machine-readable data architecture.

The concept gaining traction among product and data teams is generative engine optimization, or GEO, a counterpart to the SEO frameworks that governed digital shelf investment for the past decade. The logic is analogous: just as brands once structured their content for search algorithm ranking, they now need to structure their product data for how AI agents retrieve and prioritize options. The critical difference is that SEO operated within relatively transparent ranking signals. Agentic recommendation logic is far less legible, varies across platforms, and is evolving faster than most brand teams can track.

Retail Media in an Agentic Environment

The advertising model inside conversational agents is not simply a new format. It is a different context of attention entirely. A shopper who has submitted a specific product need to an AI agent is already past the awareness and consideration phases that most upper-funnel media is designed to generate. They have declared intent in explicit natural language. The agent’s response is the moment of influence. Sponsored placement within that response operates at the highest point of purchase intent in the funnel, though with limited transparency into how often or under what conditions that placement will surface.

Walmart’s retail media business grew 33 percent year over year in its most recent reported quarter, six times faster than total company revenue growth. That growth reflects both the strength of Walmart’s first-party data and the increasing seriousness with which brands are treating Walmart’s platform as a performance channel. The integration of sponsored placements into Sparky adds inventory that operates under different principles than keyword-based sponsored search. Brands will need to understand how their product data feeds into Sparky’s recommendation engine and whether their content is structured in ways that help rather than hinder agent-driven retrieval.

Amazon’s position in this landscape is distinct in ways that matter. Amazon’s Rufus AI shopping assistant reportedly engaged 250 million shoppers in 2025, with monthly active users growing 140 percent year over year. Shoppers who interact with Rufus are 60 percent more likely to complete a purchase, and sessions involving Rufus during the 2025 holiday period that ended in a transaction doubled relative to the preceding 30-day baseline, while non-AI sessions grew just 20 percent. These are not marginal engagement gains. They reflect a genuine shift in how discovery leads to conversion within a closed ecosystem where Amazon controls the catalog, the agent, the fulfillment promise, and the post-purchase experience.

That last point is worth sitting with. An agent can surface the right product, but the shopper still needs to trust that the order will arrive when promised and that returns will work without friction. Amazon has spent 30 years building the infrastructure and customer expectation that makes that certainty credible. An open-standard agent routing across many retailers introduces variability into the post-purchase experience that Amazon’s closed loop does not. For brands deciding where to concentrate their agentic commerce investments, that fulfillment reliability gap is not a secondary consideration.

The Supplier Readiness Gap

There is a real distance between the pace at which platforms are deploying agentic tools and the pace at which most suppliers have adapted their commercial models to work within them. A brand’s relationship with its retail partners has historically been managed through human-to-human account teams, category reviews, planogram negotiations, and promotional planning calendars. Agentic systems introduce a parallel layer of influence that operates continuously, at machine speed, and entirely outside the traditional cadence of commercial planning.

Suppliers who treat agentic commerce as a media channel problem, addressable through incremental advertising budgets, are likely to underestimate what is actually required. The readiness questions cut across functions. Product data completeness and attribute architecture are owned by category management and digital shelf teams. The content feeding agent recommendation engines is often produced by marketing agencies whose briefs were written for human-facing creative. The real-time inventory and availability signals that influence agent recommendations are supply chain and fulfillment variables. Connecting these functions around a coherent agentic strategy requires organizational coordination that most brands do not currently have in place.

There is also the customer relationship to consider. When a consumer makes a purchase through Google’s UCP-powered checkout, or through Sparky’s sponsored prompt, or through Alexa+’s contextual replenishment, the platform intermediating that transaction holds the interaction data, the purchase signal, and frequently the fulfillment relationship. Brands whose customer relationships are already mediated by retail partners now face a sharper version of a familiar problem: the consumer’s agent is making choices based on data your brand does not fully control, on a platform whose recommendation logic you cannot fully see.

The Brands That Move Now Will Have the Advantage Later

The replacement behavior showing up in consumer data does not mean traditional search is immediately irrelevant or that every purchase journey will route through an AI agent in the near term. Adoption remains concentrated among power users, and the category mix of purchases completing through agents today skews toward lower-complexity, lower-risk decisions. A ChannelEngine survey of 4,500 shoppers found that only 17 percent currently feel comfortable allowing an agent to complete a purchase autonomously.

But the trust gap is closing, and the infrastructure being built in 2026 is designed for a world where that behavior is normalized. Brands that use this period to get their product data structured, their content optimized for agent retrieval, their retail media strategies extended into conversational environments, and their supplier operations connected to real-time signals will be in a fundamentally stronger position when autonomous purchasing becomes routine.

The core question for brand leaders is not whether to engage with agentic commerce. It is whether the organizational structures and data systems built for a search-and-browse world are capable of competing in one where the shopper’s first and often only touchpoint is a prompt submitted to an AI that has never seen a banner ad, never browsed a digital shelf, and has no interest in your brand’s visual identity. What it does evaluate is accurate, structured, attribute-rich product data. That, more than any creative investment or media innovation, is what brand visibility in an agentic environment will be built on.

Conversations On Retail

Conversations On Retail is a gathering place and resource center for retail and CPG executives, built to make it easier to stay current, discover the technologies and solutions shaping the industry, and connect with the people driving it forward.

We publish news, views, and reviews from staff editors, contributing experts, and trusted partners. Some articles are developed internally, while others are submitted by industry contributors or adapted from interviews and recorded conversations with industry leaders.