"Order me 500 business cards, 350gsm, matte laminated, double-sided printing. Here's my logo." Someone says this to their AI assistant. Not to a human. To an AI.
The agent opens three online print shops, compares prices, checks delivery times, uploads the logo, configures the product, and sends the user a summary: "Printer A: 49 euros, 3 business days. Printer B: 42 euros, 5 business days. Printer C: 55 euros, 2 business days. Which should I order?"
One click. Done.
This is Agentic Commerce. And it will fundamentally change online retail. Not tomorrow. Not the day after. But sooner than most think.
What Agentic Commerce Is -- and What It Isn't
Agentic Commerce means: AI agents act on behalf of users. They don't just search, they buy. They don't just compare, they decide (or prepare decisions).
This is fundamentally different from everything we've known:
- Search engines show links. The user must click, compare, and decide themselves.
- Price comparison sites aggregate data. The user still has to order themselves.
- Chatbots on shop websites answer questions. But they don't buy.
AI agents go one step further. They understand the task, research, compare, and execute. The user delegates the entire purchase process.
Honest Assessment: What Works Today, What Doesn't?
Before imagination runs wild, here's the sober inventory:
What works TODAY
Product search and comparison: ChatGPT can search for products, compare specifications, and make recommendations. Ask "best Bluetooth headphones under 200 euros for commuters" and you'll get useful answers -- based on product data, reviews, and test reports.
Product recommendations in AI Overviews: Google already shows product recommendations in its AI Overviews -- with price, image, and link. Those with structured product data are displayed preferentially.
Specialized shopping assistants: Amazon, Shopify, and other platforms have integrated AI assistants that recommend products and simplify purchasing within their ecosystems.
What DOESN'T work yet
Autonomous checkout: No AI agent can independently complete an order and pay today. The reasons are technical and legal:
- Payment authorization requires human confirmation (PSD2, 3D Secure)
- There's no standard for agent-to-shop checkout
- Liability questions for erroneous orders are unresolved
Product configuration: Complex product configurators (color, size, material, personalization) are difficult for agents to use because most shops implement these as interactive UI elements rather than API endpoints.
Returns and complaints: The post-purchase process is agent-ready almost nowhere.
The honest statement: We're in the "search and recommend" phase, not the "autonomous shopping" phase. But the transitions are fluid, and the infrastructure is actively being built.
What "AI-Ready" Concretely Means for Online Shops
AI-Ready in e-commerce has four levels:
1. Structured Product Data (schema.org Product)
Every product needs machine-readable metadata. This sounds trivial but isn't -- most shops have product data formatted for humans, not machines:
{
"@type": "Product",
"name": "Premium Business Cards 350gsm Matte",
"description": "Double-sided printing, matte laminated",
"sku": "BC-350-MATTE",
"brand": { "@type": "Brand", "name": "PrintPro" },
"offers": {
"@type": "Offer",
"price": "49.90",
"priceCurrency": "EUR",
"availability": "InStock",
"deliveryLeadTime": { "value": 3, "unitCode": "DAY" },
"priceSpecification": {
"minQuantity": 100,
"maxQuantity": 1000
}
},
"additionalProperty": [
{ "name": "Weight", "value": "350gsm" },
{ "name": "Lamination", "value": "Matte" },
{ "name": "Printing", "value": "Double-sided" }
]
}
2. A Product API
The next step after structured HTML: a real API that AI agents can query programmatically.
GET /api/v1/products?category=business-cards&weight=350g&finish=matte&quantity=500
Shopify already provides a solid foundation with its Storefront API. WooCommerce has a REST API. Custom shops need to build this themselves. But the technical barrier is manageable.
What's critical: the API must support filtering, sorting, and availability checking. An agent that needs to load 500 products to find one is a bad agent -- and won't recommend your shop.
3. Cart and Checkout API
This is where it gets interesting -- and where we're furthest from reality. An agent would need to:
- Add product to cart
- Select configuration options (quantity, color, size)
- Set delivery address
- Choose shipping option
- Generate checkout link
Steps 1-4 are technically feasible. Step 5 is the crux: the agent generates a prepared checkout link that the user only needs to confirm and pay. This isn't an autonomous purchase -- it's a "prepared checkout" where the human makes the final decision.
Shopify supports this via the Storefront API with checkoutCreate. This is the most realistic path for the next 1-2 years.
4. Availability and Delivery Information
AI agents will prefer shops that deliver reliable real-time information:
- Is the product in stock?
- How long does delivery take?
- What shipping options are available?
- What's the free shipping threshold?
Shops that hide this information in fine print lose against shops that provide it structured and via API.
Why Shopify Merchants Have an Advantage (and Custom Shops Need to Catch Up)
Shopify bet early on API-first. The Storefront API offers:
- Product queries with filters
- Cart management
- Checkout creation
- Inventory information
- Metafields for additional product data
This means: a Shopify shop is "API-technically" already closer to AI-Ready than most custom shops. What's usually missing is documentation for agents and discovery mechanisms (agents.json).
Custom shops -- whether on WooCommerce, Magento, or proprietary builds -- need to build this API layer first. It's not rocket science, but it requires planning and deliberate architecture.
The Immediate Benefit: What AI-Ready Delivers TODAY
Better Visibility in Google Shopping and AI Overviews
Structured product data following schema.org is the foundation for Google Shopping, rich snippets, and AI Overviews. Those who mark up their products machine-readably appear more frequently and prominently in search results.
Higher Conversion Through Better Product Data
Working on structured data forces you to complete and standardize product information. Full specifications, clear availability indicators, precise delivery times -- this improves not just AI readability but also human purchase decisions.
Product Comparisons by AI as a Marketing Channel
When ChatGPT users ask for product recommendations, products with complete, structured data are preferred. This is a new marketing channel that requires neither cost-per-click nor ad budgets -- just good data.
Foundation for Your Own AI Features
Those with structured product data can build their own AI features: an intelligent product finder ("I'm looking for a gift for a 40-year-old who loves cooking"), personalized recommendations, or automated product descriptions.
The Biggest Mistakes Online Shops Make
Mistake 1: Product Data Only for Humans
"Our favorite hoodie -- cuddly soft, in 5 colors." That's marketing. For AI, what's missing: material (80% cotton, 20% polyester), sizes (S-XXL), color codes, care instructions, weight. The emotional description can stay -- but it needs a structured counterpart.
Mistake 2: Prices and Availability Only on the Product Page
When an agent wants to compare 50 products, it has to load 50 pages. An API that delivers price and availability for a product list in one request is orders of magnitude more efficient.
Mistake 3: Configuration Options as JavaScript Widgets
Size selection via dropdown, color selection by clicking a square, quantity via plus/minus button. Intuitive for humans. Unusable for agents. The solution: offer the same options as API parameters too.
Mistake 4: No Bulk Queries
"Show me all running shoes under 150 euros in size 43." This requires a filter API. Most shops have a faceted filter on the website -- but no API offering the same functionality.
What the Path to an AI-Ready Online Shop Looks Like
Phase 1: Structured Product Data (2-3 weeks)
- schema.org Product markup for all products
- Standardize specifications (size, color, material, weight)
- Set up Google Merchant Center and validate feed
- Complete product data (identify and fill gaps)
Phase 2: Product API (3-4 weeks)
- Build REST API for product queries (or configure Storefront API)
- Filter endpoints (category, price, availability, properties)
- Implement pagination and rate limiting
- Create API documentation
Phase 3: Cart Integration (2-3 weeks)
- Cart API with configuration options
- Generate prepared checkout links
- Provide shipping and delivery information via API
Phase 4: Agent Discovery (1 week)
- Create agents.json
- Publish API documentation for agents
- Health and status endpoints
The timeline varies significantly depending on existing tech stack. A Shopify shop can implement phases 1 and 2 in considerably less time because the infrastructure already exists.
The Elephant in the Room: Payment Authorization
The biggest open question in Agentic Commerce is payment. No serious expert claims that AI agents will independently execute payments in the near future. The regulatory hurdles are high (PSD2, Strong Customer Authentication), liability questions unresolved.
The most realistic model for the coming years:
- Agent configures the order
- Agent generates a checkout link
- User reviews and confirms
- User authenticates payment (fingerprint, Face ID, TAN)
This isn't autonomous shopping. It's "assisted checkout" -- the agent does the work, the human pushes the button. But even this saves enormous time and reduces cart abandonment.
Why Major Platforms Will Set the Rules
Amazon, Shopify, and Google will set the standards. Amazon already has an AI shopping assistant (Rufus). Shopify is investing heavily in AI features for merchants. Google Shopping is increasingly AI-driven.
For small and medium online shops, this means: those not visible on these platforms must build their own AI-Ready infrastructure. That's more effort, but it also offers independence. And structured product data is the foundation in any case -- whether for ChatGPT, Google AI Overviews, or your own AI product finder.
Conclusion: The Product API Is the New Storefront
In brick-and-mortar retail, the shop window was the most important sales factor. In early e-commerce, it was the product photo. Then Google placement. In the next phase, it will be the product API.
Not because people will stop visiting shops. But because between the user and the shop, there's increasingly an AI agent. And this agent decides based on data, not on pretty pictures.
The technology for AI-Ready e-commerce exists. The standards are defined (schema.org, Storefront APIs, agents.json). What's missing is implementation by the majority of shops.
Investing now builds the infrastructure for tomorrow -- while simultaneously improving Google ranking, product data quality, and today's conversion rate. That's not a bet. That's solid craftsmanship.
