Selling outside the US? A guide to where AI shopping assistants actually work for CPG brands

Read enough AI shopping announcements and you start skimming the last paragraph, which is where the geography lives. "Available to U.S. customers." "Rolling out in the U.S." "Live for ChatGPT users in the U.S." For a brand that only sells in America, that line doesn't matter. For the many CPG brands that also ship to Canada, sell through a UK grocer or have a distributor in Scandinavia, it decides which half of the AEO advice applies to them.

This guide goes market by market. It covers what is live, where it's live, and what a food, beauty or household brand should do in each place. We used the companies' own announcements wherever we could and dated everything, because these lines move. Our news log tracks the launches as they land.

The short version

Shopping features in AI assistants are mostly an American thing for now. The assistants themselves are not. Someone in Manchester can ask ChatGPT which oat milk foams best and get an answer with brand names in it. What they don't get yet is the shopping layer: product cards fed by merchant catalogs, carts and checkout.

That split gives you a simple rule. The work that shapes what an assistant says about your product works everywhere. The work that shapes how it sells your product only pays off where a shopping feature exists.

What's live, and where

Table: AI shopping features and where they are available, from each company's announcement (checked October 27, 2026)

Feature	Company	Where it is available	What it does for a shopper
Shopping in ChatGPT	OpenAI	US	Product results from merchant feeds; purchase on the merchant's site or app
Universal Cart	Google	US (launched Oct 15, 2026)	One cart across Search and the Gemini app; Google Pay or transfer to the store
UCP integration hub in Merchant Center	Google	Rolling out in the US; Australia and Canada next year	Merchant onboarding for checkout in AI Mode and Gemini
Alexa for Shopping	Amazon	US (May 2026)	Questions in Amazon's search bar, comparisons, Buy for Me on other stores
Perplexity shopping	Perplexity	US at launch (Nov 2025)	Product cards and PayPal checkout
Copilot Checkout	Microsoft	US at launch (Jan 2026)	Purchase inside Copilot; merchant stays merchant of record
Muse	Meta	US and Canada	A personal agent that shops on the user's behalf
AI shopping assistant	Prisjakt	Sweden, Norway, Denmark, Finland	Conversational product comparison on Prisjakt's price data
AI shopping agent	Naver	South Korea	Recommendation and comparison; payment pilot in some categories

Two caveats. The Perplexity and Copilot rows reflect what each company said at launch, and either may have widened since; we didn't find a later announcement that says so. And "US" for Amazon means the US store. Amazon runs separate stores in other countries, and its announcement covers only the American one.

The pattern in that table

Every big American platform started at home. The two that went further were Meta, which added Canada, and Google, which has named Australia and Canada as next for merchant onboarding. Zuckerberg's explanation for keeping Muse in two countries, as reported by Payments Dive, was cost: the computing infrastructure is still too expensive for a wider release. That's a useful hint that expansion depends on money and capacity, which means nobody can promise you a date.

The US: the full stack

If the US is one of your markets, it's where every lever is live at once. We have covered most of these in detail elsewhere, so this is the checklist rather than the explanation.

• ChatGPT. OpenAI's merchant page says Shopify and Etsy catalogs are already integrated, and other merchants apply for a waitlist. It also says purchases happen on the merchant's own site or app. Our notes from the ChatGPT shopping feed piece still apply, and there's a short version in the ChatGPT shopping tip.
• Google. Universal Cart and checkout in AI Mode both run on Merchant Center data. A clean product feed and the newer attributes matter more than they did a year ago. We wrote up the conversational attributes and what Universal Cart means for your add-to-cart numbers.
• Amazon. Alexa for Shopping answers questions from your listing, so the work is on the listing itself. Our Alexa for Shopping playbook has the steps.
• Copilot and Perplexity. Both use checkout partners (Shopify, PayPal and Stripe for Copilot, PayPal for Perplexity). If you're on Shopify, check your admin for Copilot Checkout, because Microsoft said Shopify merchants would be enrolled automatically after an opt-out window.
• Muse. There's no feed to submit. Muse uses your normal storefront, so your product pages, stock and checkout have to work for a machine as well as a person.

Canada: Muse now, Google next year

Canada is the one market outside the US with a major shopping agent already running. Meta told investors that about 10 million people had signed up for Muse across the US and Canada. It didn't split that number by country, so don't assume a share.

Because Muse works through ordinary storefronts, the practical work for a Canadian store is the same as for an American one. A few points need extra care north of the border:

Bilingual product pages

If your Canadian site has English and French versions, make sure both carry the same facts. An agent that lands on the French page and finds a different serving count or an older price is no better off than one that finds a typo. This is the matching-facts rule with a second language added.

Prices and currency

Show the Canadian dollar price clearly on the page itself, not just in a currency switcher that loads after the page renders. An agent reading the page should never have to guess which currency it's looking at.

Getting ready for Google's hub

Google's Merchant Center help page says the UCP integration hub, which handles onboarding for checkout in AI Mode and Gemini, is rolling out in the US first, "with Australia and Canada to follow next year." Its advice for preparing is plain: make your Merchant Center data "as robust and up to date as possible." Canadian brands can do that now, and it helps ordinary Shopping listings in the meantime.

The Nordics: a comparison site went first

The most interesting launch this week wasn't American. On October 26, Prisjakt, the Nordic price comparison service, released a conversational shopping assistant. Shoppers describe what they need, and it recommends and compares products using Prisjakt's own data: prices, price history, expert reviews and user reviews.

The scale behind it is real. Prisjakt says its database covers more than 18 million products from over 12,800 retailers across eleven markets. The assistant itself is narrower for now. It's available to signed-in users in Sweden, Norway, Denmark and Finland.

Prisjakt contrasts its assistant with general-purpose AI services and says it's built on the company's own continuously updated product and price data. For a CPG brand, that means the assistant is only as good as the listings and reviews Prisjakt already holds for your products. So in the Nordics the job looks a lot like ordinary retail hygiene:

1. Search Prisjakt for your own products in each country and check that they appear under the right name, with the right size and a current image.
2. Ask the retailers who carry you whether their feeds reach Prisjakt, and fix gaps in the product data they send.
3. Read the reviews attached to your products there. An assistant that quotes reviews will quote the bad ones too.

Our tip on retailer listings covers the general case. Prisjakt is a clear example of it.

South Korea: Naver's agent is growing fast

Naver launched its AI shopping agent in March and, according to an Aju Press report on October 25, is now piloting payment inside the agent for some product categories, with plans to roll it out by the end of the year. It's also moving from the app and mobile web to desktop.

Naver gave growth figures rather than raw numbers.

Show the data

Measure	Times the launch level
Monthly active users	5x
Daily conversations	5x
Daily transaction volume	4x

Source: Naver, reported by Aju Press, October 25, 2026. Naver described the first two as about fivefold and the third as quadrupled. It did not publish user or transaction counts, so these are multiples of an unknown base.

Two other details are worth a CPG team's attention. Digital and electronics made up 40% of the agent's transaction volume, so groceries and beauty are a smaller slice for now. And 78% of transactions came from Naver Plus members, which means the agent is mostly being used by people who already shop on Naver.

The example query Naver gave is a good one for food brands: a request for a post-workout protein powder at around 70,000 won. That's two conditions, use and price, in one sentence. Naver said queries with two or more conditions are becoming more common, and that it plans to rely more on blog posts, community content and purchase reviews to explain its recommendations. If you sell in Korea, your Naver listing needs the attributes that answer those conditions, and reviews on Naver carry real weight.

The UK, the EU, Australia and everywhere else

Here there is, for now, no major shopping layer to optimize for. That doesn't mean there's nothing to do. Shoppers in these markets already ask assistants product questions, and the answers name brands. Those answers come from what the model learned in training and from pages it fetches when it searches. Neither depends on a shopping feature launching in your country.

So the work is the part of AEO that has nothing to do with feeds:

• Let the crawlers in. Check robots.txt and your CDN's bot settings so that search crawlers such as OAI-SearchBot can fetch your pages. Our crawler tip has the list.
• Keep facts consistent across countries. If the UK pack is 400g and the US pack is 14 oz, say so explicitly on each regional page. A model that sees both numbers without context may print the wrong one.
• Earn local coverage. Answers often cite roundups and reviews. A "best oat milk" list in a UK newspaper will be cited for UK prompts in a way your US press won't.
• Watch your retailer listings. Tesco, Ocado, Coles or Carrefour product pages can be what an assistant finds when it searches for your product in that country, so they need the same facts as your own site.

When shopping features do reach these markets, brands that did this work first will have less to fix. Google has named Australia for next year. OpenAI says it plans "to expand to more merchants and regions over time" without saying where.

How to check what shoppers in each market see

This is the step most brands skip, because it's tedious. Ask the questions your buyers ask in each country, in the local language, and record which brands show up and what the answer says about yours. Use the shopper's phrasing, not your marketing copy. A Swedish shopper asks about "havredryck", not "oat beverage". Repeat the same questions monthly so you can see whether anything changes.

We do this by hand for a small set of prompts. Some brands buy tools for it instead. XENA Intelligence is one company in that space: it asks assistants such as ChatGPT, Perplexity, Gemini and Amazon's assistant shopper-style questions and scores products on how they show up, and its SearchPanel tool shows the actual answers for each SKU. Whichever way you do it, run the same check in each market rather than assuming the US results apply everywhere.

A plan by market

Table: Where to spend AEO effort by market, based on what is live as of October 27, 2026

Market	Live AI shopping features	Do first	Add when relevant
United States	ChatGPT, Google, Amazon, Perplexity, Copilot, Muse	Feeds, Merchant Center attributes, Amazon listings, storefront that agents can use	Checkout settings per platform
Canada	Muse	Storefront facts in English and French, clear CAD pricing	Merchant Center prep for Google's hub next year
Sweden, Norway, Denmark, Finland	Prisjakt assistant	Retailer feeds reaching Prisjakt, reviews	Brand-site basics
South Korea	Naver agent	Naver listing attributes and reviews	Payment pilot categories as they open
UK, EU, Australia, others	None of the above yet	Crawler access, consistent regional facts, local press	Feeds when platforms expand

If you only have time for one thing outside the US this quarter, make it the facts. Check that every regional page and every retailer listing describes the same product the same way, with the local size, price and currency written plainly. It costs little, it works in every market, and it's the base everything else builds on when the shopping features arrive.