Retail AI Report: Agentic Commerce is Now Operational Infrastructure

Google Completes Its Agentic Commerce Stack and Confirms Live Deployment at Retail's Biggest Names

Across Google I/O on May 19 and Google Marketing Live on May 20, Google confirmed that its Universal Commerce Protocol checkout is now operational at Nike, Sephora, Target, Ulta Beauty, Walmart, Wayfair, and select Shopify merchants, enabling consumers to complete purchases directly from Google Search, Gemini, and YouTube with Google Pay. The underlying infrastructure powering this is Universal Cart, a persistent AI-powered shopping hub that accumulates a consumer's intent signals and product selections across sessions and surfaces, paired with the Agent Payments Protocol (AP2), which enables AI agents to transact on behalf of users within user-defined spending limits and removes the human from the final conversion step entirely. Affirm and Klarna buy-now-pay-later options are now embedded in Google Pay for these transactions, and new Merchant Center tools will allow brands to track performance specifically within AI-generated search results. Google also expanded UCP to Canada and Australia, with the UK next.

Why This Matters: Google has moved the point of sale upstream, ahead of the retailer's own website, into the discovery layer it controls. The architectural ambition is significant: Google is positioning itself as the connective tissue between consumer intent and retail fulfillment across Search, Gemini, YouTube, and Gmail, which is to say, most of the internet. For retailers already integrated with UCP, this means access to streamlined conversion at the moment of peak intent. For those who are not, the risk is concrete: AI agents routing purchases on behalf of millions of users will preferentially complete transactions where the infrastructure exists to do so. Being absent from UCP is not a neutral choice; it is a decision to cede high-intent, AI-mediated traffic to competitors who are present. The BNPL integration with Affirm and Klarna further reduces friction for high-AOV categories, which is where UCP's incremental revenue impact will be most pronounced. Executives should be accelerating integration timelines and auditing product data quality, because agents select on structured signals, not brand equity.

Source: TechCrunch / Google Blog 

Source: Google Blog / Digital Commerce 360

Walmart's Sparky Is Now the Industry's AI Benchmark

Walmart's Q1 fiscal 2027 earnings call delivered the retail industry's most concrete AI performance data to date: weekly active users of its Sparky AI shopping agent more than doubled in a single quarter, and customers who use Sparky spend 35% more per order than those who don't. CEO John Furner used the phrase "becoming AI native," as a description of how the company is restructuring operations, from supply chain intelligence to in-store associate support. Sparky has expanded to Spanish-language support, personalized replenishment, and meal planning, and Walmart recorded its ninth consecutive quarter of U.S. e-commerce growth above 20%.

Why This Matters: The 35% average order value lift is the number boards and investors will fixate on, and that creates a board-level pressure point for every retailer that hasn't yet deployed a comparable capability. More important than the headline figure is what's driving it: AI-assisted discovery and replenishment naturally expand basket size because they surface relevant adjacent products that human shoppers browsing a catalog wouldn't encounter. Retailers who dismiss Sparky's results as "Walmart scale" miss the point: the underlying dynamic applies at any scale, and the performance gap will widen with each quarter of inaction.

Source: Digital Commerce 360 / PYMNTS

Klarna Plants Its Flag Between ChatGPT and Every Merchant's Checkout

On May 20, Klarna launched its Shopping Search app inside ChatGPT, connecting more than 100 million products and 400 million merchant listings across 13 markets directly to ChatGPT's conversational interface. Consumers describe what they want in natural language, receive visual results with live pricing and availability from multiple merchants, and are redirected to retailer sites to complete purchase. Organic results are ranked by relevance; sponsored placements are available and clearly labeled. The technical foundation is Klarna's Product Search MCP server, where MCP stands for Model Context Protocol, an emerging standard that lets AI models query external databases in real time rather than relying solely on their training data.

Why This Matters: The strategic implication is stark: Klarna has inserted itself as the default product discovery and monetization layer between ChatGPT's estimated 700 million monthly users and the merchants those users might otherwise have reached directly. For retail brands, this creates a new "AI shelf," analogous to the digital shelf in traditional e-commerce SEO, where visibility depends on feed quality, pricing competitiveness, and relevance signals rather than ad spend alone. Every brand that has a paid search strategy and a Google Shopping feed now needs an equivalent strategy for how it appears inside AI-native discovery surfaces. The brands that move first to optimize for these channels will establish positions that are difficult and expensive for slower movers to displace.

Source: PYMNTS / Digital Commerce 360 / Business Wire

‍

Retail Media's Blind Spot: AI Agents Don't See Your Banner Ads

A PYMNTS analysis published May 22 laid out the structural threat that agentic commerce poses to retail media networks, the high-margin advertising businesses that have become critical profit centers for retailers like Amazon, Walmart, and Target. Retail media is built on the fundamental assumption that human shoppers, browsing visually, can be influenced by placement, creative, and urgency signals. AI agents shopping on behalf of users operate on an entirely different logic, optimizing on price, delivery performance, availability, and explicit user preferences. Sponsored banners and premium shelf placements are, to an AI agent, invisible noise.

Why This Matters: The consequence is a potential structural erosion of retail media revenue models that currently generate billions of dollars annually and carry margins significantly higher than merchandise sales. Industry analysts are beginning to use the term "AI Engine Optimization" (AEO), structuring product data, fulfillment signals, and API integrations so that AI agents select your products, as the successor discipline to traditional retail media buying. This doesn't mean retail media disappears overnight; human shoppers remain the majority of traffic for now. But executives managing retail media P&Ls need to be running scenario analyses on what a 20%, 30%, or 40% shift toward agent-mediated transactions does to network revenue, and CPG brand partners asking the same question will demand answers. The window to architect a credible response is narrowing.

‍

Source: PYMNTS

Fanatics and Vertex AI: What "Good Search" Actually Looks Like in a High-Emotion Catalog

Fanatics, the sports merchandise giant, announced a partnership with Google Cloud's Vertex AI to overhaul its on-site search, fixing a problem that had been quietly costing the company sales for years: fans searching for their team's current starting lineup were receiving results featuring retired players or irrelevant merchandise. The new system understands the contextual language of sports fandom, including team affiliations, jersey types, and event-specific lines, and handles catalog volatility during high-stakes moments like March Madness or MLB's City Connect jersey releases, when thousands of SKUs are added rapidly and fan intent is hyperspecific and time-sensitive.

Why This Matters: The Fanatics case is instructive precisely because the problem seems simple. Fans searching for a live player should find that player's jersey. That this required a purpose-built AI overhaul reflects how poorly traditional keyword-based search handles catalog complexity, emotional context, and real-time inventory changes simultaneously. Any retailer with a large, dynamic, or emotionally resonant catalog, whether fashion, beauty, home goods, or specialty food, faces a version of this same problem. The business case is direct: poor search results don't just frustrate customers, they redirect purchase intent to a competitor who surfaces the right product faster. AI-powered contextual search is not a technology upgrade; it is a revenue recovery project.

Source: The AI Marketers Newsletter

What to Watch

Google's UCP expansion to the UK is the next concrete milestone, as that market will test whether the agentic checkout model holds in a more privacy-sensitive regulatory environment and whether European consumer adoption differs meaningfully from U.S. patterns. Separately, watch for CPG brand partners of major retail media networks to begin demanding AEO metrics and agent-visibility audits in their media buying negotiations. The first public friction between a major CPG and a retail media operator over AI-era attribution will signal the shift from theoretical concern to commercial reckoning. And as Walmart's Sparky benchmarks circulate through boardrooms, expect Q2 earnings season to produce a wave of AI shopping agent announcements from retailers working to demonstrate comparable capability. Parsing signal from noise in those announcements will be the editorial challenge of the summer.