Cart is not merely a convenience feature. It is the structural centrepiece of Google’s pivot to what the industry is calling “agentic commerce” – a model in which AI doesn’t just assist the decision-making process, it is the decision-making process. Built on top of Google’s Shopping Graph, a machine learning-powered database of more than 60 billion product listings, the Universal Cart allows consumers to accumulate items across every Google-owned surface – Search, Gemini, YouTube, Gmail – into a single, persistent, intelligent cart. The system then does something no checkout widget has done before: it works autonomously in the background, scanning for better prices, flagging product incompatibilities, surfacing loyalty perks and monitoring stock availability across Google’s entire search index. The initial rollout, set to begin in the United States this northern summer, will include some of the world’s most recognisable retail brands: Nike, Sephora, Walmart, Target, Ulta Beauty and Wayfair, alongside select Shopify merchants including Fenty and Steve Madden. Google’s 24/7 AI personal assistant, Gemini Spark – announced alongside the Universal Cart – is designed to integrate directly with the cart infrastructure, paving the way for a system that can research, shortlist, and complete a purchase entirely without human input. That is the true definition of an AI shopping agent, and it is closer than most brands are prepared for. Google’s billion-query gambit The strategic logic is clear and aggressive. Google handles approximately one billion shopping queries per day, and until now, it has never owned a single transaction. The Universal Cart changes that calculus entirely. Where Google Search historically acted as the top of a funnel it did not control, the Universal Cart pulls the entire funnel – discovery, comparison, cart management, price monitoring, checkout – inside Google-owned real estate. The comparison to Amazon’s “Buy for Me” feature, launched earlier last year, is inevitable, but the distinction matters enormously. Amazon operates a closed commercial ecosystem while Google operates the internet. For brands and marketers, the implications extend well beyond which platform hosts the “Buy Now” button. The past year has already required a significant recalibration, as businesses adapted their digital presence for both traditional SEO and the emerging practice of AI Optimisation (AIO) – ensuring products surface favourably across AI-driven discovery surfaces, not just keyword-ranked search results. The Universal Cart accelerates that urgency dramatically. When an agentic AI is making the first round of decisions on a consumer’s behalf, brand equity must be legible to the machine before it ever becomes visible to the human. The hard data questions The data dimension of this shift may prove to be the most consequential of all. The Universal Cart draws on a consumer’s activity and preferences across Search, Gemini, YouTube and Gmail – a behavioural portrait of unprecedented granularity. Combined with Google Wallet’s existing visibility into payment preferences, loyalty memberships and merchant offers, Google will accumulate a real-time, deeply personalised commercial profile of hundreds of millions of shoppers. Brands have historically owned this intelligence; it underpins seasonal buying decisions, product development and pricing strategy. Google’s terms and conditions for its Universal Commerce Protocol state that brands “remain the Merchant of Record, maintaining full control over customer relationships and data” – but the company is yet to clarify whether this protection carries over to the new Universal Cart architecture. That ambiguity, for any serious retail executive, should read as a red flag. Operationally, the Universal Cart raises the bar for backend readiness in ways that will separate prepared retailers from exposed ones. Because the agentic AI continuously cross-references live inventory and pricing across Google’s index, any gap between a brand’s actual stock position and its digital product feed becomes a customer-facing failure – instantly. Retailers with clean, dynamic Merchant Center data, competitive pricing infrastructure, and real-time inventory synchronisation will gain disproportionate visibility. Those lagging on product data hygiene will find themselves algorithmically deprioritised before a human being ever enters the picture. Too convenient to resist  What Google is building, piece by piece, is a commerce operating system – one that sits between brands and their customers in a way that is extraordinarily difficult to disintermediate. The convenience of a single, intelligent cart that already knows consumers’ payment methods, loyalty points, style preferences and price sensitivity is not a feature most will opt out of. Instead, most people are expected to take the trade. For brands, the question is no longer whether to engage with this new paradigm. It is whether they understand, yet, what they stand to lose if they don’t. If agents are doing the buying, the brand touchpoint isn’t the campaign. It isn’t the product page. It’s the data signal – the feed, the price point, the stock flag – that tells an AI your product is worth choosing. The most important marketing decision of the next five years may not be creative at all. It may be a metadata field. Further reading: Inside Google’s big bet to get businesses using its AI agents