Artificial intelligence is rapidly reshaping how people shop online, and with the release of its new ‘Universal Commerce Protocol,’ Google is making a clear play to sit at the center of that transformation.
At the opening of this year’s National Retail Federation (NRF) conference, Google unveiled a sweeping vision for what it calls the “agentic commerce” era — a future in which AI systems don’t just recommend products or answer customer questions, but actively carry out tasks on behalf of shoppers and retailers alike.
Alongside new AI-powered retail tools announced through Google Cloud, the company also introduced a new open-source framework, the Universal Commerce Protocol (UCP). Google hopes this protocol will become an industry standard for how AI agents handle everything from product discovery and checkout to post-purchase customer support.
The promise is significant: smoother shopping journeys, more personalized experiences, and less technical complexity for retailers. But as Google tightens its grip on AI-driven commerce, the announcements also raise important questions about control, competition, and the growing dependence of retailers on a handful of powerful platforms.
From browsing to “doing”
While shopping online has traditionally centered around browsing for products, the increasing use of LLMs has led to the creation of more conversational, action-oriented experiences — where AI assistants help consumers find the right product, build a cart, place an order, and resolve issues later — all in one flow.
This is increasingly referred to as “agentic commerce,” where AI agents are designed to plan, reason, and complete multi-step actions autonomously (with user permission).
To support this shift, Google Cloud has announced Gemini Enterprise for Customer Experience (CX) — a platform that unifies shopping, customer service, and support into a single AI-driven system. Powered by Google’s most advanced Gemini models, the platform includes pre-built agents that can be deployed quickly and customized by retailers.
Google says these agents can do far more than simply answer questions about products. They can act as digital concierges, processing text, voice, and images to help customers discover products, assemble orders, and even handle post-purchase problems. According to Google, retailers including Kroger, Lowe’s, Papa John’s, and Woolworths are already using this system.
The goal, the company says, is to turn “passive browsing into active doing” — and to make every interaction feel tailored to the individual shopper.
The Universal Commerce Protocol: standardization or consolidation?
The most strategic announcement at NRF, however, may have been the launch of the Universal Commerce Protocol.
UCP is designed to act as a shared language for AI agents across the retail journey, spanning search, product discovery, checkout, and support. Google says the protocol will allow retailers to plug into AI-powered commerce without having to build and maintain complex integrations between different systems.
Speaking to CNBC about the new protocol, Google’s vice president of ads and commerce, Vidhya Srinivasan, said “it’s very important to have a standardized way [of implementing AI ecommerce], so we can scale these things and everyone can be prepared for all the various steps to happen.”
Google says UCP has been co-developed with retailers and platforms including Shopify, Etsy, Wayfair, and Target, and that it will soon power a new checkout experience inside Google’s AI Mode and Gemini app. Payments will initially run through Google Wallet, with support for alternatives like PayPal planned for the future.
On paper, this sounds like a win for retailers — particularly smaller ones that lack the resources to build sophisticated AI systems. A common protocol could reduce friction, speed up innovation, and make it easier to adopt agentic commerce tools.
But there’s an obvious tension here. By defining the standard, Google also positions itself as the gatekeeper.
If UCP becomes widely adopted, retailers may find that a growing share of discovery, checkout, and customer interaction happens on Google-controlled surfaces — from AI chat interfaces to search results — rather than on their own websites. That raises familiar concerns about disintermediation, loss of brand control, and increased dependence on Google’s ecosystem.
A crowded and competitive AI commerce landscape
Google’s push comes amid intense competition in AI-driven shopping.
OpenAI has already announced Instant Checkout inside ChatGPT, which allows users to purchase certain products directly within the chat interface. OpenAI’s own Agentic Commerce Protocol, developed with Stripe, is also open source — and could emerge as a rival standard to Google’s UCP.
Perplexity is taking a similar approach, partnering with PayPal to let users buy products, book travel, and purchase tickets without leaving its app. And Amazon continues to blur the lines between marketplace and agentic shopping, with features like “Buy for Me,” which lets AI purchase items from other retailers on a shopper’s behalf.
According to a report from McKinsey, AI-powered retail could represent a $3 trillion to $5 trillion global opportunity by 2030 — a figure that explains why so many tech giants are racing to control the starting point of the shopping journey.
For retailers, this competition may lead to more than a bit of head-scratching. More platforms offering agentic commerce tools means more innovation — but also more fragmentation, and more strategic decisions to be made about where to show up, which standards to support, and how much control to give away.
Advertising, reimagined for AI
Unsurprisingly, Google’s announcements also tie directly into its core business: advertising.
As shopping shifts into conversational interfaces, Google is testing new ad formats designed for AI-driven discovery. One example is a new “Direct Offers” format, which would allow retailers to surface targeted discounts when a user expresses purchase intent inside Google’s AI Mode chatbot.
For retailers, this could create new opportunities to reach high-intent shoppers at precisely the right moment. But it also reinforces a long-standing concern: that visibility and success in AI-driven commerce may increasingly depend on paid placement rather than organic discovery. If AI agents become the primary interface between consumers and products, the rules of ecommerce SEO, branding, and customer acquisition could change dramatically — and not necessarily in ways that favor smaller or independent retailers.
Early wins — and early warning signs
Google showcased a range of early success stories at NRF, from Authentic Brands Group using AI-generated creative to boost ad performance, to The Home Depot rolling out an in-store “Magic Apron” agent that helps customers plan projects and find products down to the aisle level.
These examples highlight the upside of agentic commerce: faster workflows, better personalization, and more seamless experiences across online and physical retail.
But the potential downsides are just as real.
Handing more of the customer journey to AI agents introduces new risks around errors, bias, and accountability. If an AI agent builds the wrong cart, applies the wrong discount, or mishandles a support issue, who is responsible — the retailer, the platform, or the AI provider?
There are also data concerns. Agentic commerce relies on deep access to customer behavior, preferences, and transaction data. Retailers may gain insights, but they may also be sharing increasingly valuable information with platform providers whose incentives (or values) don’t always align with theirs.
And finally, there’s the risk of dependency. As Google, OpenAI, Amazon, and others race to define the infrastructure of AI-driven shopping, online retailers face a familiar dilemma: adopt early and risk lock-in, or wait and risk falling behind.
Why this matters to ecommerce merchants
For ecommerce merchants, Google’s move into agentic commerce brings both opportunity and risk. On the positive side, tools like the Universal Commerce Protocol could make it much easier — and cheaper — for online retailers to provide AI-driven, conversational shopping experiences.
The trade-off is control. As more discovery, checkout, and support happens on Google’s platforms, merchants may lose direct access to customers, data, and brand differentiation, while becoming ever more dependent on Google’s standards and advertising tools.
Get ecommerce news you can use
Sign up below and we’ll send you email digests of all the latest news and developments in the world of ecommerce, tips on how to grow an online business, and special offers.
Chris Singleton is the Founder and Director of Ecommercetrix.
Since graduating from Trinity College Dublin in 1999, Chris has advised many businesses on how to grow their operations via a strong online presence, and now he shares his experience and expertise through his articles on the Ecommercetrix website.
Chris started his career as a data analyst for Irish marketing company Precision Marketing Information; since then he has worked on digital projects for a wide range of well-known organizations including Cancer Research UK, Hackney Council, Data Ireland, and Prescription PR. He then went on to found the popular business apps review site Style Factory, followed by Ecommercetrix.
He is also the author of a book on SEO for beginners, Super Simple SEO.