Agentic Commerce: What your online shop needs for AI agents to shop with you

Your next customer no longer types “black sneakers size 43” into Google. They tell ChatGPT: “Find me comfortable black sneakers under 120 euros, ideally with fast delivery and free returns.” Three seconds later, the agent has compared three shops, checked reviews and prepared the purchase. The question is: was your shop one of them?

Agentic commerce describes shopping by AI agents that research on behalf of customers, compare products and, when authorized, buy directly. The human sets the goal. The agent handles the rest.

This is already the present: in March 2026, Shopify activated Agentic Storefronts for all 5.6 million shops. Shopware founded the Agentic Commerce Alliance. And Google, OpenAI and Perplexity are in an open race over whose agents shop best.

Search volume for “agentic commerce” was 1,900 queries per month in January 2026. Twelve months earlier it was practically zero. McKinsey projects that by 2030, AI agents will drive one trillion US dollars in retail revenue in the US alone, and up to five trillion worldwide. A recent IBM study found that 45 percent of consumers already use AI when shopping.

What this means: the human no longer searches and decides alone. The AI searches, filters and prepares the decision. If you run a shop and are not found, you are out before the potential buyer even hears about your product.

60 percent of respondents can imagine AI support when shopping, but only 9 percent would delegate the purchase entirely. The human still makes the final call. But the AI decides which options land on the table in the first place. For e-commerce shops and their owners, that changes the rules at a point many still underestimate: visibility before the actual search.

The three dominant platforms take different routes toward a similar goal.

ChatGPT Shopping is the best-known entry point. OpenAI introduced shopping features in spring 2025, tested Instant Checkout in the fall and pulled a first version again in March 2026. Too simple for production use. The new direction: dedicated retailer apps that let shops plug directly into the ChatGPT purchase flow. With 880 million monthly active users, this is no longer a fringe channel. In many categories, its volume already exceeds Google Shopping.

Perplexity combines product search with source citations and offers paying subscribers a “Buy with Pro” feature: purchase directly without switching shops. Perplexity is especially strong for information-heavy purchase decisions, because research and checkout flow into each other.

Google AI Mode builds on the existing Shopping Graph. Google knows product data, prices and availability from millions of shops and uses Gemini not only to deliver purchase recommendations but also to justify them. For online shops that already use Google Merchant Center and structured data, this is a head start.

What all three platforms have in common: they do not read product pages like a human. They process structured data, schema markup, product attributes and reviews. A nicely designed shop without a clean data foundation is invisible to agents. As an AI agency that implements e-commerce projects technically, we see this every day: technical cleanliness is no longer optional, it is the baseline for the agent channel.

For AI agents to shop in stores reliably, they need standardized ways to communicate. Two protocols dominate the discussion.

Universal Commerce Protocol (UCP): Google and Shopify presented UCP at the NRF trade show in January 2026. The protocol defines how AI agents query product data, availability, prices and checkout processes through an open interface. More than 20 partners are already on board: Walmart, Target, Visa, Zalando. UCP is open, API-based and designed to connect discovery and transaction.

Agentic Commerce Protocol (ACP): OpenAI and Stripe developed ACP with a focus on the checkout flow. Anyone buying through ChatGPT goes through an ACP-secured payment process. The reference implementation is documented at agenticcommerce.dev. ACP focuses more on closing the sale and less on the search before it.

The difference in practice: UCP covers the entire purchase journey, from the first request to the transaction. ACP starts at checkout. The two protocols do not exclude each other.

What this means for shop owners: if you support no protocol, agents will not find you reliably or cannot enable autonomous purchases. This is not a technical detail for the IT department, but a decision about how open your shop is to the next sales channel.

Shopware and Shopify are both actively integrating the protocols into their platforms. For Shopware shops and Shopify shops, that is good news: if you run your platform cleanly, you benefit from the work the platform vendors invest.

Shopware has responded strategically: in February 2026, Shopware founded the Agentic Commerce Alliance — a European alliance of retailers, technology partners and service providers. Its goal is to promote open standards for agent-driven commerce and bring a European perspective to the global protocol discussion.

Of practical relevance for Shopware merchants is above all the Shopware AI Copilot, available in the commercial editions. The Copilot helps create product descriptions and content. That directly helps agents find a better data foundation.

What Shopware shops need to do right now:

• Implement structured data. Product schema, Offer schema and FAQPage schema are a must. Without these markups, agents cannot reliably extract product data.

• Open API endpoints for catalog data. Shopware 6 has a powerful REST API that must be configured correctly and be accessible so external agents can query product data.

• Clean up product data. Complete attributes, correct categorization, consistent units. What earns a poor feed score in Google Shopping also blocks AI agents.

As a certified Shopware agency, we put these steps into practice in everyday project work — from schema markup in the theme to API configuration.

With the launch of Agentic Storefronts on March 24, 2026, Shopify made a clear move: all 5.6 million Shopify shops were automatically opened to AI agents. No opt-in required, only an opt-out for those who decline. Shopify treats the agent channel as the default, not as an experiment.

Shopify Catalog makes product data from Shopify shops directly accessible to AI agents, and also to shops on other platforms such as BigCommerce that can join in.

Knowledge Base app: With this app, merchants store their brand voice, frequently asked questions, return policies and other information that gives agents context when retrieving your products. What is stored here influences how ChatGPT or Perplexity describe your shop and whether customers see your offer as trustworthy.

Agentic Plan: Shopify offers a new pricing tier for brands that only want to make their product catalog available to agents, without running a full shop. This is especially interesting for B2B manufacturers that are not building a D2C channel but still want to appear in agent flows.

As an official Shopify Plus Partner, we advise on how to set up these features optimally: from the Knowledge Base to the Catalog integration.

API-first architectures have a structural advantage in the agentic commerce discussion: if you built your shop headless, you have already laid the foundation for the agent channel.

In headless setups, content is not hidden in templates but available through structured APIs. Storyblok as a headless CMS delivers content as JSON, a format AI agents can process natively. Vue Storefront as a decoupled frontend cleanly separates presentation from the data layer, giving agents direct access to product data.

commercetools has also announced its own Agentic Commerce Suite. API-first architecture becomes the infrastructure for the agent channel.

For growing headless e-commerce shops, now is the right time to review the existing API architecture for agent compatibility and close gaps before the channel delivers meaningful traffic.

Agents do not read colors and do not judge layout. But design remains decisive, in a different way than before.

What agents actually read: schema markup, product descriptions, review texts, FAQ data, return and shipping rules, certifications. Everything structured and machine-readable feeds into the agent’s decision. The brand voice in the Knowledge Base (that is, how your brand is described in AI answers) depends directly on the quality of your structured content. That is corporate design in the age of AI: not only visual, but semantic.

Product reviews, delivery times, return rate, security certificates: what used to be a trust signal for human visitors is now a data point for agents. A shop without reviews is hard for an agent to assess and, when in doubt, will not be recommended.

When an agent presents three options, the human makes the final choice. And that is where the visual experience comes back: with comparable offers, the quality of the shop experience at checkout often decides. Component-based UX & service design is therefore not a nice-to-have but the conversion lever after the agent has made its shortlist.

One more technical point that is often underestimated in practice: many agent implementations abort the purchase process when checkout hangs, error messages appear or mandatory registration interrupts the flow. A cleanly built checkout without forced registration and with fast load times is the minimum requirement for the agent channel. Core Web Vitals are not only a Google ranking factor, they are a quality signal for completed purchases.

1. Schema markup implemented?
Product, Offer and FAQPage schema are a must. Without these markups, agents cannot reliably extract product data. Google’s Rich Results Test shows you what your shop currently delivers.

2. Structured product data complete?
Title, description, price, availability, EAN/GTIN, brand, category, variants: is everything filled in completely and correctly? Missing attributes mean agents cannot classify your product correctly in comparisons.

3. Product feed up to date and error-free?
If you already run a Google Shopping feed, you have a good foundation. This feed must be updated daily and be error-free, because agents use the same data structure.

4. FAQ data stored in structured form?
Are your FAQs marked up as FAQPage schema? Common questions about shipping, returns, product details and payment methods are valuable context for agents when preparing a purchase.

5. Return and shipping rules machine-readable?
Agents actively check return periods and shipping costs. If this information only exists as body text on a subpage, you are not delivering structured data. Markup and clear copy are both needed.

6. Checkout without mandatory registration?
Guest checkout must be possible. Agents buying on behalf of a customer cannot create accounts. A registration requirement blocks the purchase entirely.

7. Brand voice maintained in the Knowledge Base?
For Shopify shops: fill in the Knowledge Base app with brand voice, frequently asked questions and unique selling points. What is stored there influences how agents describe your brand.

8. Reviews available and integrated?
Product reviews are trust signals for agents. Shops without reviews lose out against shops that provide reliable review data. Reviews must also be marked up with structured data.

9. Shopify: Agentic Storefronts configured?
Default activation is not enough. Are Knowledge Base and Catalog set up correctly? Is product data complete? These are the three points that decide the quality of the agent channel.

10. Shopware: API endpoints for catalog data open?
Is the Shopware 6 API configured and reachable? No authentication hurdles for public catalog data? Already using the Shopware AI Copilot for product copy?

Agentic commerce is not a trend that becomes relevant in two years. It is already here, in Shopify’s default settings, in ChatGPT’s shopping features, in Google’s AI Mode. The question is not whether AI agents will shop in stores, but whether your shop is part of it.

If you already rely on structured data, complete product data and a working checkout, you are closer to agent readiness than you think. No complete system migration, no big budgets. Priorities in the right places.

We combine e-commerce consulting with technical implementation: Shopware, Shopify, headless. We know which adjustments make the biggest difference. Let’s talk about your shop. We will show you where you stand and what the next step is.