Google I/O 2026: What the Agentic Commerce Announcements Mean for Brands
Last Updated:
At Google I/O 2026 on May 19, Google announced Universal Cart, expanded AI Mode capabilities, and a new Merchant Center attribute schema called Conversational Attributes. Alongside the new launches, Google also gave updates on two existing pieces of agentic infrastructure: the Universal Commerce Protocol (UCP) and the Agent Payments Protocol (AP2).
Taken together, this is Google laying out the working model for a shopping ecosystem where AI agents handle discovery, comparison and checkout on the shopper's behalf, across Search, Gemini, YouTube and Gmail. Brands will influence those agents through the structured data they expose, not through on-site merchandising or paid placements at the point of decision.
We spoke to Vogue about what this means for fashion specifically. The mechanics apply just as much to beauty, electronics, home, grocery and travel.
The high-level shift
Three things are happening at once across the announcements:
Discovery moves from keywords to intent. Shoppers describe full requirements, and agents reason across structured product data to pick winners. The product feed becomes the front door.
The point of sale moves inside Google. Universal Cart and UCP let shoppers check out without leaving Search, Gemini, YouTube or Gmail. The brand stays the merchant of record, but loses the on-site moments that historically influenced the final decision.
Some purchases stop involving the shopper at all. Under AP2 mandates, agents transact autonomously against pre-set rules. The brand encoded into the rule wins by default.
The combined effect is that the commercial discipline that wins in 2027 is agentic commerce optimisation: making the brand legible, accurate and recommendable to the agents reading product data across surfaces. It sits alongside SEO and retail media rather than inside either of them.
Here's the breakdown of what was announced, what it means in practice, and what to do about it.
1. Universal Cart and the new reasoning layer
What's new
A shopping cart that works across merchants and across Google surfaces (Search, Gemini, YouTube, Gmail)
Runs in the background finding deals, monitoring price drops, alerting on restocks
Reasons about the basket itself, flagging incompatibilities (Google's example was custom PC parts) and suggesting alternatives
Built on Google Wallet, so it factors in payment perks, loyalty status and merchant offers
Checkout via Google Pay in a few taps, with launch partners including Nike, Sephora, Target, Ulta Beauty, Walmart, Wayfair, and Shopify merchants such as Fenty and Steve Madden
The brand remains the merchant of record
Rolling out across Search and Gemini in the U.S. this summer, with YouTube and Gmail to follow
Implication
The cart needs structured attributes to do its job. When it checks compatibility, fit, materials, ingredients, dimensions or care instructions, it is reading those fields directly from Merchant Center.
Products with incomplete or inconsistent attributes get filtered out of the cart's reasoning before they reach the shopper. This applies across the catalogue, not just hero SKUs, because the cart reasons across the whole basket.
What to do
Audit attribute coverage across the full catalogue. Mid-tail and long-tail SKUs need the same coverage as bestsellers, because one missing field can disqualify a multi-product combination.
Re-classify attribute fields as commercial, not operational. Compatibility, fit, dimensions, ingredients and care notes now drive selection. Treat them with the same scrutiny as price and stock.
Reconcile attribute data across Merchant Center, the website, and retailer feeds. Conflicts between sources are a common reason products drop out of recommendation logic.
2. Conversational Attributes in Merchant Center
What's new
A new Merchant Center attribute schema built for AI surfaces like AI Mode
Designed to answer the longer, more nuanced questions shoppers ask in conversational interfaces
Complements the existing feed rather than replacing it
Reuses product data brands already maintain for PIM systems and retailer integrations
Implication
The standard Google Shopping feed was built for keyword discovery: title, description, GTIN, category, price. Agents reading those fields in AI Mode and Universal Cart need more than that to answer questions like "is this suitable for dry, sensitive skin" or "will this fit a 6ft 2in frame with a long inseam."
Most brands already have this data. It sits in the PIM, in the website copy, in retailer-specific feeds, and in Amazon listings. The work is to push it into Merchant Center in the format Google has specified, and to keep it consistent with what's on Amazon, Walmart and Target, because the same agents are reading across all of those surfaces.
What to do
Map Conversational Attributes against the existing PIM. The raw data is usually already there, scattered across systems.
Apply Amazon-listing rigour to the Merchant Center feed. It is now the canonical source for AI surfaces.
Set a quarterly maintenance cadence. Attributes drift as ranges evolve, and stale data produces confidently wrong agent answers, which is worse than missing data.
3. AI Mode, Search agents and generative UI
What's new
From Google's I/O 2026 Search announcement:
Gemini 3.5 Flash is the new default model in AI Mode globally
The Search box has been rebuilt to handle longer queries and accept text, images, files, videos or Chrome tabs as input
Information agents monitor the web 24/7 for restocks, new drops, price changes and any criteria a shopper sets, then notify them when conditions are met
Generative UI lets Search build custom dashboards and "mini apps" on the fly, which shoppers return to over time
AI Mode has surpassed 1 billion monthly users, with queries more than doubling every quarter since launch
Implication
Two practical changes:
Queries are intent-shaped, not keyword-shaped. Shoppers describe full requirements ("a coffee table under 120cm wide, solid wood, ships before the end of the month, matches a mid-century sofa"). The data needed to win these queries goes beyond what wins on standard search.
Information agents make binary decisions. When an agent is monitoring for "sustainable running shoes under £150 with a wide toe box," the product either matches the spec or it doesn't. There is no second chance through retargeting or ad spend.
What to do
Build out the long-tail attribute layer. Certifications, materials, sustainability credentials, sizing tolerances, performance specs.
Stress-test products in AI Mode directly. Run the nuanced queries that match the brand's strengths. If flagship products don't appear, the gaps in the brief are visible immediately. Without ongoing visibility tracking across AI surfaces, this kind of testing is a one-off snapshot rather than a feedback loop.
Identify which monitoring queries the brand naturally wins. Restock alerts, new arrivals and price drops are agent-triggered moments. The brand with the cleanest stock and price feeds gets the notification surface.
4. UCP and AP2: updates to the agentic plumbing
What's new
The Universal Commerce Protocol (UCP) and the Agent Payments Protocol (AP2) were both launched before I/O. At I/O 2026, Google gave updates on both:
UCP is the open standard Google co-developed with retailers as a common language for agents and merchants. New tech partners have joined recently to help steer the standard.
UCP-powered checkout is expanding from the U.S. to Canada and Australia in the coming months, with the U.K. to follow
UCP is coming to YouTube in the U.S., and to new verticals starting with hotel booking and local food delivery
AP2 lets agents make secure payments on a shopper's behalf, with guardrails on brands, products and budget
AP2 uses tamper-proof digital mandates that create a verifiable record of every transaction
AP2 is now rolling out across Google products, starting with Gemini Spark
Implication
When checkout happens inside Google's cart, the brand loses these levers:
Creative at the point of sale
Merchandising and cross-sell
Last-minute upsell
Email capture
On-site retargeting setup
The only remaining lever is whether the agent surfaced the brand in the first place.
AP2 extends this further. Under standing mandates, agents can transact autonomously without the shopper opening a brand's website at all.
What to do
Make UCP integration a near-term priority in every live market. Treat it the same way brands treated Buy on Google or Shopify integrations when those launched.
Audit what the brand actually loses under agentic checkout. Some on-site moments were genuinely valuable (style guidance, fit advice, gifting curation). Move those upstream into the data and content the agent reads. Others were friction the shopper will be glad to skip.
Plan for autonomous purchases under AP2. If a household tells an agent to "reorder our usual coffee," the brand encoded into the mandate wins. Loyalty becomes a data structure rather than a marketing programme.
5. The new brand-customer asymmetry
What's new
Universal Cart pulls together:
Cart additions across Search, Gemini, YouTube and Gmail
Purchase history (for the proactive recommendations layer)
Cards, loyalty and offers via Wallet
Standing mandates via AP2
Gmail, Google Photos and soon Google Calendar via Personal Intelligence in AI Mode, now expanding to nearly 200 countries and 98 languages
Implication
The personalisation that makes Universal Cart useful runs on data the brand selling the product never sees. Google understands the shopper in higher resolution than the brand making the sale.
Closeness to the customer used to be a competitive advantage brands built through email lists, accounts, apps and loyalty programmes. In an agentic world, that closeness is largely rented from the platform.
What to do
Invest harder in first-party data and direct relationships. Shoppers who choose to engage directly with a brand are now a self-selected high-value segment.
Push owned data assets upstream. Loyalty data, purchase history and preference data should feed agent understanding through clean attribute structures, retailer integrations and signed-in experiences, rather than sitting in a CRM no agent can read.
Plan around the asymmetry rather than denying it. Strategies built on the assumption that the brand still owns the relationship will misallocate budget.
6. Privacy, aggregation, and the trust question
What's new
Universal Cart, Personal Intelligence, AP2 and Wallet aggregate signals across nearly every Google product. Google has built in transparency and granular controls, including selective Gmail or Photos connection to AI Mode.
Implication
The features work because of the aggregation. Pulled into one interface, Search, Gmail, YouTube, Wallet and Photos create a more detailed picture of a shopper than has existed before.
Most shoppers will accept the trade because the experience is better. Regulators in the EU and U.K. are likely to revisit what is acceptable to pool in one interface, so the current data flow should not be assumed permanent.
What to do
Build strategy on the assumption that today's data flow may narrow. The DMA and equivalent frameworks elsewhere will shape what Google can do over time. Plans that depend on current access being permanent are fragile.
Use the trust gap as a positioning lever. Brands that handle shopper data visibly well, and offer genuine value in exchange for direct engagement, have a clear differentiator against the aggregator.
7. What to do this quarter
In two to three years, a meaningful share of repeat and considered purchases will be set by instruction rather than browsed. The shopper tells an agent what they want, the agent executes against it, and the brand finds out it was selected when the order lands. Mood, story and creative still matter, but upstream of the agent rather than at the point of sale.
Five priorities for the next 90 days
Run a Conversational Attributes gap analysis against the Merchant Center feed
Stress-test top-selling SKUs in AI Mode with realistic, nuanced shopper prompts, and track visibility across surfaces over time rather than as a one-off
Confirm UCP integration status with the e-commerce platform and key retailer partners
Separate genuine value from friction in on-site moments, and move the valuable parts upstream into the data, content and off-page signals (reviews, Reddit, forums, editorial) that the agents actually read
Set a 12-month roadmap for agentic commerce optimisation as a named, resourced discipline with a clear owner, rather than an extension of SEO
The brands that get this work done in 2026 will be the ones the agents recommend in 2027. The ones that wait will be selecting from a smaller set of remaining levers.
How Azoma helps
Azoma is the agentic commerce optimisation platform built for exactly this shift. Three things in one place:
Listing optimisation at scale across Amazon, Walmart, Target and the broader agentic surface, so the structured data and copy that Universal Cart, AI Mode and Conversational Attributes read from is built for how agents actually reason
Off-page PR and Reddit programmes that put brands into the reviews, threads, forums and editorial sources the models trust when they decide who to recommend
An AI visibility platform that tracks end-to-end how brands surface across AI responses, so you can see exactly which queries you're winning, which you're losing, and what changed week to week
If agentic commerce is on the roadmap for 2026, this is the stack to get in place before competitors do. Book a demo and we'll show you what your visibility looks like today, where the biggest gaps are, and the quickest path to closing them.
Article Author: Max Sinclair