What Agentic Commerce means for e-commerce teams, how OpenAI and Shopify are shaping the new buying flow, and what stores should do before AI-driven shopping becomes standard.
We already wrote about ChatGPT Shopping, where AI recommends products. That article is still relevant, but the situation has changed dramatically in just a few weeks.
OpenAI no longer just recommends products. AI agents now independently search, compare, and choose the best products for a specific user request. The purchase still happens on the seller’s site, but the entire selection process is handled by AI. This is called Agentic Commerce, and it is no longer a concept. It is a working protocol.
In this article we will break down what Agentic Commerce is, how the new protocol works, why Shopify and Walmart already connected to it, and most importantly, what Ukrainian businesses need to do so they do not become invisible to AI buyers.
What Agentic Commerce is
Agentic Commerce is an e-commerce model where an AI agent acts on behalf of the buyer at every stage of the funnel, from product search to checkout. It does not just suggest, it does the work.
Imagine telling ChatGPT: “I need running shoes for asphalt, budget up to UAH 5,000, delivery in Kyiv.” The AI agent:
- Scans the catalogs of connected stores
- Filters by price, availability, and reviews
- Compares the 3-5 best options with photos, prices, pros, and cons
- Shows a side-by-side comparison
- Redirects to the seller’s site for checkout
The difference from the earlier ChatGPT Shopping is fundamental. Before, AI only showed links. Now it actively searches, compares, and recommends the best option for a specific request. The final purchase still happens on the seller’s site.
This is not theory. According to Shopify Q3 2025 data, AI traffic to Shopify stores grew 7x since January 2025. AI orders grew 11x. And this is only the beginning, because the Agentic Commerce Protocol was launched just recently.
Agentic Commerce is not a new version of search. It is a new buyer. An AI agent with a budget, criteria, and authority to buy.
How the Agentic Commerce Protocol works
Agentic Commerce Protocol (ACP) is OpenAI’s technical standard for how AI agents interact with product catalogs. In simple terms, it is the language AI uses to talk to stores.
Product feed → AI catalog ▼
The store provides a structured product catalog: names, descriptions, prices, photos, availability, and reviews. This can happen through a direct integration such as Shopify Catalog or through a standardized feed.
AI agent receives the user request ▼
The user describes what they need in natural language, with context, budget, and preferences. The AI agent interprets the request and builds the search criteria.
Discovery: search across catalogs ▼
The agent scans connected catalogs, filters by criteria, and ranks by relevance. Now it does that with visual product cards and side-by-side comparisons directly in chat.
Recommendation with explanation ▼
AI shows the top options and explains why it chose them. Not just a list, but an argumented choice for a specific request. The user sees photos, prices, key features, and reviews.
Redirect to the seller’s site ▼
After choosing a product, ChatGPT redirects the user to the seller’s site for checkout. OpenAI initially tested Instant Checkout, meaning purchase inside chat, but later moved away from that model because of low conversion and integration complexity. The current model is simple: AI finds it, the seller sells it.
Who has already connected
Shopify Catalog – integration went live on March 24, 2026. All 5.6 million Shopify stores automatically received “Agentic Storefronts,” AI-optimized storefronts that ChatGPT, Microsoft Copilot, Google AI Mode, and Gemini can read and show to buyers.
Walmart launched integration with ChatGPT for product search and recommendations. Users find products through AI and then go to Walmart to complete the checkout, while Walmart keeps control of checkout, loyalty, and payments.
Etsy was the first partner in the Instant Checkout tests, but OpenAI later moved away from that model because of low conversion. Etsy now works on a discovery plus redirect model.
These are not experiments. These are production systems from the world’s largest platforms. The model is clear: AI finds the best product, and the seller closes the deal on their own site.
Why this matters for Ukrainian e-commerce
You might think, “Walmart, Shopify, that is all about the US. It is irrelevant for my store in Ukraine.” That is a mistake that can be expensive.
Buyer behavior is already changing
Gartner predicted that by 2026 traditional search volume would drop by 25%. We can already see it: more people start product search not with Google, but with ChatGPT or Perplexity. Not because they are lazy, but because AI gives a personalized answer instead of 10 ad links.
Ukrainian users are no exception. ChatGPT is available in Ukraine, subscriptions are growing, and the habit of asking AI is formed day by day.
Invisibility is the new problem
In Google you can be on page 3 and still get found through ads or long-tail keywords. In AI commerce, you are either in the recommendation or you do not exist. There is no middle ground.
If the AI agent scans catalogs and does not find your product because there is no structured data, because the bot is blocked, or because the description is just a list of specs, it recommends a competitor. There is no second chance.
Early entry equals advantage
Agentic Commerce is just starting. Most Ukrainian stores do not even know what it is yet. Those who prepare now will get first-mover advantage. When AI traffic becomes mainstream, they will already be in the catalogs.
It is the same situation we had with GEO optimization: the ones who started earlier collected the gains.
What to do right now: 5 steps
Step 1: Add structured data, Product schema
This is the foundation. Without schema.org/Product markup, the AI agent cannot extract product information such as name, price, availability, or rating. It sees text, but it does not understand that it is a product.
Minimum:
name– product namedescription– natural language descriptionimage– a quality photooffers.price+offers.priceCurrencyoffers.availability– availabilitybrand.nameaggregateRating– rating
We gave a detailed guide with a JSON-LD template in the ChatGPT Shopping article. Use it as the base.
Check: the Google Rich Results Test will show whether your products are marked up correctly.
Step 2: Optimize product descriptions for AI
The AI agent is looking for an answer to the buyer’s question, not a list of features. Your descriptions should sound like a recommendation from a real person.
Bad: “XYZ coffee machine. 15 bar pressure. 1.2 L tank. Automatic milk frother. Black.”
Good: “XYZ coffee machine makes espresso and cappuccino with one touch. 15 bar pressure is enough for a rich crema. The 1.2 liter tank makes 8-10 cups without refilling. The automatic frother creates milk foam like a cafe.”
The second version can be quoted by AI as a recommendation. The first one will be ignored because it does not answer any buyer question.
Step 3: Connect to platforms with AI integrations
If you are on Shopify, that is already good. Agentic Storefronts are rolling out automatically. Your catalog becomes available to AI agents without extra work.
If you are on WooCommerce or another platform:
- Allow OAI-SearchBot and ChatGPT-User in robots.txt, and see the detailed instructions in our ChatGPT Shopping article
- Submit the product feed directly to OpenAI when that option becomes available for more platforms
- Use XML or JSON feeds with current data: prices, availability, photos, and ratings
If you sell on marketplaces such as Rozetka or Prom, watch for their AI integrations. Large marketplaces will connect too. It is only a matter of time.
Step 4: Collect and structure reviews
The AI agent analyzes reviews as arguments for a recommendation. “All good, 5 stars” is not an argument. “Bought it for my wife’s birthday, she does not know how to make coffee, figured it out in 5 minutes, now she drinks latte every morning” is exactly what AI will quote.
How to get useful reviews:
- Ask after purchase: “Who was it for? What did you like most? What surprised you?”
- Mark up reviews with schema.org/Review
- Respond to negative reviews constructively. AI sees the seller’s reaction too
Step 5: Track AI referral traffic in GA4
If you do not measure it, you do not manage it. Create a GA4 segment:
- Source =
chatgpt.comorchat.openai.com - Medium =
referral
Track sessions, conversion rate, average order value, and time on site. Compare it with organic and paid. If AI traffic is growing, you are on the right path. If it is zero, something is blocked or not marked up.
More on analytics setup: our GA4 + GTM guide.
E-commerce readiness checklist for AI commerce
PDF: 20-point checklist from robots.txt to reviews. Go through it and see whether the AI agent can actually see your store. Ready to print.
Get it on Telegram →Agentic Commerce vs Google Shopping vs marketplace
Three sales channels, three different models. Here is how they compare:
| Parameter | Agentic Commerce | Google Shopping | Marketplace (Rozetka, Prom) |
|---|---|---|---|
| How the product is found | AI agent scans catalogs automatically | User searches by keywords | User searches inside the platform |
| Placement cost | Free for now | CPC ads or free listings | Commission on sales, 5-20% |
| Personalization | Deep – context, dialogue, memory | Minimal – filters and sorting | Basic – browsing history |
| Brand control | Medium – AI decides what to show | High – you control ads | Low – standardized template |
| Purchase | Redirect to seller’s site | Go to seller’s site | On the marketplace platform |
| AI readiness | Critical success factor | Secondary for now | Depends on the platform |
| Who makes the decision | AI agent recommends, human confirms | Human chooses from search results | Human chooses from the catalog |
Main conclusion: Agentic Commerce is not a replacement for Google Shopping or marketplaces. It is a new channel with a fundamentally different mechanism. In Google Shopping you pay for attention. In a marketplace you pay for the transaction. In Agentic Commerce you invest in data quality and AI recommends you for free.
The smart strategy is to be in all three. Compare it with how Performance Max works for Google Shopping and build a multi-channel approach.
Want to prepare your store for Agentic Commerce?
I will audit your AI visibility: schema markup, robots.txt, description quality, reviews, and readiness for AI catalogs. Message on Telegram · My services
Frequently asked questions about Agentic Commerce
Agentic Commerce is an online commerce model where an AI agent, such as ChatGPT, independently searches, compares, and buys products on behalf of the buyer. Instead of manually scrolling catalogs and comparing prices, the person describes what they need, and AI does everything else, right down to checkout.
ChatGPT Shopping is product recommendations inside chat. Agentic Commerce is the next level: the AI agent does not just recommend, it acts autonomously. It has a protocol for direct connection to store catalogs, visual cards with side-by-side comparisons, and deep analysis for a specific request. The purchase still happens on the seller’s site.
Partly. AI search and product recommendations through ChatGPT work globally if your site is open to OAI-SearchBot. The purchase happens on the seller’s site, so there are no geographic limits for discovery. The Agentic Commerce Protocol is an open standard that will expand. It is worth preparing now.
As of today, no. OpenAI does not sell ad placements and does not take a commission from sales through ChatGPT Shopping. Recommendations are based on product quality, structured data, and reviews. But this may change. OpenAI is already testing partner integrations, and monetization through ads in the future is possible.
Three first steps: 1) check robots.txt and allow OAI-SearchBot and ChatGPT-User, 2) add schema.org/Product markup through a plugin such as Rank Math SEO or Yoast SEO Premium, 3) rewrite product descriptions so they answer buyer questions instead of listing specs. That is the minimum needed to make the store visible to AI agents today.
Conclusion: the businesses that prepare will win
Agentic Commerce is not far in the future. Shopify, Walmart, and Etsy are already working with AI agents. OpenAI launched the protocol. ChatGPT is already shaping product discovery, comparison, and purchase decisions for users.
For Ukrainian e-commerce this means one thing: a new sales channel has appeared, and you are either visible or invisible. And this channel does not require an ad budget. It requires quality: quality data, quality descriptions, and quality reviews.
Your plan for this week:
- Check robots.txt to see whether OAI-SearchBot can read your catalog
- Check schema.org/Product on product pages
- Rewrite the 10 most popular descriptions, from specs to recommendation-style answers
- Create an AI traffic segment in GA4
- Download our AI commerce readiness checklist and go through all 20 points
Businesses that do this now will get traffic and sales when Agentic Commerce becomes mainstream. Everyone else will have to catch up.
Read also: ChatGPT Shopping: how to get into AI recommendations · GEO optimization for AI search · Google Ads from scratch · About me · Case studies · Services
Comments
Discuss on Telegram