Shopify, VTEX and agentic commerce: what each platform already gives AI agents
By Daniel Pisano, Partner & CGO at Vurdere
Shopify and VTEX both connect stores to AI agents through Google's Universal Commerce Protocol (UCP), in different ways. Shopify publishes a catalogue that agents query and uses rating and review count as a quality signal. VTEX has integrated UCP with Gemini in the US. On both platforms, agent checkout is still limited to the US market.
Why does the platform matter in agentic commerce?
Very few merchants will implement an agentic commerce protocol themselves. The platform does it. It decides which product data leaves the store, in what format and for which assistants. So the practical question for a merchant is not "which protocol will win" but "what does my platform already send to agents, and what is still up to me".
I went through the public documentation of two platforms that come up constantly in my conversations with retailers. The picture is quite different.
What does Shopify already give agents?
Shopify has built the most complete and best-documented setup. It has three parts:
Shopify Catalog organises product information and sends it to connected AI platforms. Price and inventory changes in the admin flow through automatically.
Agentic Storefronts is a sales channel in the admin that connects the store to ChatGPT, Microsoft Copilot, AI Mode in Google Search and the Gemini app, with no apps or custom integrations. The merchant stays the merchant of record.
The Agentic plan lets brands on other platforms add products to Shopify Catalog without replatforming. Shopify says it rolled out globally in March 2026.
Underneath sits UCP, which Shopify co-developed with Google. The numbers Shopify shared for Q1 2026 show the scale: AI-driven traffic to Shopify stores grew 8x year over year, and orders from AI-powered searches grew nearly 13x. The catch for merchants in Europe and elsewhere: according to Shopify, purchasing through Agentic Storefronts is limited to US buyers.
Where do reviews fit on Shopify?
This is the part I care about most. Shopify's help centre says one of the listing quality signals measures a product's average rating and review count. It adds a condition: only reviews verified by trusted sources count. That quality score affects how AI channels may rank products.
In the developer documentation, the catalogue search tool accepts a rating filter with two thresholds: a minimum rating on a 0 to 5 scale and a minimum number of reviews. An agent can ask for "only products rated 4 or above with at least 50 reviews". A product with no reviews does not even enter that search.
Where does that rating come from? Shopify has two standard metafields for it: reviews.rating, the average rating, and reviews.rating_count, the total number of ratings. If a store's review app does not write to those fields, the rating sits outside the place the platform set aside for it.
An independent test published on 16 October 2026 by No Hacks backs this up. The author ran 1,305 searches against Shopify Catalog and analysed 20,410 product rows. In the top three positions, 66% of products had more reviews than the middle product of their list; in positions 26 to 50, only 40% did. Star rating made little difference. One finding matters for European sellers: a buyer located in the UK or Portugal shared only about half of a US buyer's results. The author is clear that this is a correlation, measured from outside, on a single day.
What does VTEX already deliver?
At VTEX Day 2026, held on 16 and 17 April in São Paulo, VTEX announced its integration with Google's UCP, enabling discovery and checkout directly in Gemini and AI Mode with native cart sync. The announced availability is in the US.
VTEX also introduced AI Personal Shopper, which combines semantic search and conversational recommendations on the storefront, and AI Workspace, with four pre-trained agents for catalogue management, promotions, search optimisation and business insights. In a live demo, the search optimisation agent generated estimated incremental revenue of about R$80,000 (Brazilian reais) a month for one client.
What I could not find: public VTEX documentation stating which product fields reach the agent through the UCP integration, or whether review ratings are among them. That does not mean they are missing. It means VTEX merchants should ask.
How do the two platforms compare today?
Area | Shopify | VTEX |
Protocol | UCP, co-developed with Google | Integration with Google's UCP |
Assistants | ChatGPT, Copilot, AI Mode and Gemini | Gemini and AI Mode |
Agent checkout | US buyers | Announced for the US |
Reviews | Documented quality signal; filter by rating and count | No public documentation found |
On-site AI | Not the focus of this article | AI Personal Shopper and AI Workspace agents |
What should merchants outside the US do now?
Agent checkout has not reached shoppers in Europe through these platforms yet. That leaves time to prepare, and almost all of the work is about data:
Check your product categories. Shopify excludes products it classifies as uncategorised from the catalogue.
Put dimensions, materials and compatibility in the description, as text. Agents read text.
On Shopify, confirm that your review app writes to reviews.rating and reviews.rating_count.
On VTEX, ask your account manager which data reaches agents through UCP and whether review ratings are included.
Grow your volume of verified reviews. It is the strongest lever in the evidence available today.
How does Vurdere help on both platforms?
The Vurdere app for Shopify writes the aggregate rating and review count to the native metafields reviews.rating and reviews.rating_count, the same fields Shopify defines as standard. Vurdere also works with VTEX, Magento and other platforms, with installation in about 3 hours and one week of validation.
The rest is volume. In A/B tests with current customers, Vurdere collected up to 10x more reviews and measured up to 3x conversion rate. In our study of more than 120,000 consumers in Brazil, review requests sent by WhatsApp converted at up to 15%, against 5% for email.
My view
Shopify is ahead, and not because its protocol is better. It is ahead because it has documented what it wants. When a platform writes that verified rating and review count feed its quality score, merchants know where to put the work.
VTEX has a real advantage that tends to get overlooked: agents inside the store itself, like Personal Shopper. For retailers whose customers will wait a while for agent checkout, that may pay off sooner. Still, I would like to see VTEX publish what reaches external agents. You cannot optimise what you cannot see.
One warning for both camps: the No Hacks test suggests review count matters more than star rating. Anyone who has spent years protecting a 4.9 average built on 12 reviews should rethink that priority.
Frequently asked questions
Can Shopify merchants sell through ChatGPT?
Yes. Through Agentic Storefronts, Shopify stores can sell in ChatGPT, Copilot, AI Mode and Gemini. Purchasing through this channel is limited to US buyers.
Does VTEX support agentic commerce?
In April 2026 VTEX announced an integration with Google's UCP, with discovery and checkout in Gemini and AI Mode, available in the US. It also offers on-site agents such as AI Personal Shopper.
Do reviews affect product ranking in Shopify Catalog?
Shopify says average rating and verified review count form a quality signal that affects ranking in AI channels. An independent test found a link between more reviews and higher positions.
What are the reviews.rating and reviews.rating_count metafields?
They are Shopify's standard fields for storing a product's or variant's average rating and total number of ratings. Review apps can write the rating and count to them.
Can a European store sell to AI agents yet?
Through the integrations covered here, agent checkout is still limited to the US. Preparing your catalogue and reviews is worth doing now.
Comments