Reading Time: 10 Minutes

Agentic Commerce: How AI Is Revolutionizing Online Shopping and What We've Built With "James"

Imagine watching a YouTube video about the latest Apple products and being able to buy any product shown right away. Or tell a digital concierge what you’re looking for, and they’ll find the best deals from thousands of online stores in seconds. That’s exactly what Agentic Commerce is.
But let’s start from the beginning :).

What is Agentic Commerce?

Until now, online shopping has worked like this: you visit an online shop, search for items yourself, compare prices manually and place an order. Agentic Commerce turns this principle on its head. Instead of you going to the online shops, they come to you – facilitated by an intelligent AI agent.

Shopify, the world’s largest e-commerce platform, has created an open protocol for this: the Universal Commerce Protocol (UCP). It is a kind of common language that enables AI agents to communicate with millions of online shops – securely, in a standardised way and in real time.

What exactly can an AI agent do in agentic commerce?

What makes it special is that it works across multiple shops. A single agent can find, compare and order products from hundreds of different online shops.

The technology behind it: MCP

Behind the scenes, Shopify uses what are known as MCP (Model Context Protocol) servers. These servers act as the interface between AI agents and the Shopify commerce infrastructure. Developers can connect their AI applications to Shopify’s vast network via MCP – and gain access to millions of products with real-time prices and availability.

James: Our Concierge Shopping Service

Based on this technology, dkd has developed James: a digital concierge for personalised online shopping. James connects to the entire Shopify ecosystem via the Shopify Catalog API and offers two innovative ways to discover products.

Below, we use two specific examples to demonstrate how James puts the capabilities of Agentic Commerce into practice. The first use case shows how context-based AI product search delivers better results, whilst the second demonstrates how YouTube videos are transformed into interactive shopping experiences through automatic product recognition.

Use Case 1: Intelligent product search with context

The problem
Traditional product searches aren’t particularly helpful. If you search for ‘laptop’ here, for example, you’ll get thousands of results with no context whatsoever. But what you really need is an advisor who understands your context.

The solution: Context-based search
With James, you don’t just describe what you’re looking for, but also why. The system offers over 20 predefined context templates – from ‘Birthday present for a woman’ to ‘Sustainable & eco-friendly’ and ‘Tech enthusiast’.

Alternatively, you can write your own text, for example as follows:

»"I'm looking for a high-quality laptop for video editing, with a maximum budget of €2,000"«

What makes searching in James so special?

The product search in James goes far beyond traditional shop or marketplace search functions. Rather than simply displaying lists of results, James combines contextual understanding, intelligent filters and AI-powered processing to quickly highlight the most relevant products from a wide selection.

• Context templates
  20+ scenarios such as gift occasions, lifestyle or budget
• Extensive filters
  Price range, country of dispatch, category, availability, condition
• Across all shops
  A search scans thousands of shops at the same time
• AI-generated highlights
  Each product is presented with a summary and its unique selling points
   

Product details and price comparison

Click on a product to see all the details at a glance: specifications, key features, reviews and a direct checkout button.

Use Case 2: Video Shopping – From Watching to Buying

Das Problem
You're watching a tech review on YouTube. The host shows you 15 different products. You want to buy one of them—but which one was it again? And where can you find it at the best price? Normally, you'd have to do the research yourself.

The Solution: Automatic Video Shopping
James analyzes the transcript of a YouTube video and automatically detects when shoppable products are mentioned. As you watch the video, relevant products appear on the right—in real time and in sync with the video.
 

How does YouTube Video Shopping with AI work?

To turn a regular video into an interactive shopping experience, James combines transcript analysis, product recognition, and automatic store search. This ensures that relevant products appear on screen precisely when they are mentioned or featured in the video.

1. Load video:
   Enter a YouTube URL (or use the default demo video)
2. Transcript analysis:
   James uploads the transcript and analyzes every single section
3. Product ID:
   A score-based system determines whether a section is product-relevant—based on brands, prices, specifications, and product categories
4. Automatic search:
   A product search is triggered only for relevant sections
5. Real-time display:
   The products update as the video plays

AI-powered product recognition in video: Not every sentence is relevant to a purchase

To ensure that users see only truly relevant product recommendations, James evaluates each section of the transcript individually. The system determines whether a specific product is actually mentioned—or whether the text consists solely of general conversation, farewells, or sponsor mentions.

"Sticky" products: Always something to see

When the presenter is talking about something general (transitions, summaries), the products do not simply disappear. Instead, the products from the last relevant section remain visible – with a discreet indication of where they come from.

The timeline in detail

In the transcript timeline, you can see at a glance:

• Gold stars for product categories
• Text highlighted in sections relevant to the product
• Text in a lighter font for filler content (introductions, sponsors, transitions)
• The product counter shows how many products were found in a segment

A comparison of agentic commerce: product search and video shopping with AI

With James, there are two different approaches to product recommendations — depending on whether users already have a specific purchase in mind or are simply looking for inspiration. Both approaches use AI to find relevant products more quickly, compare them across shops and make them available for immediate purchase.

Both routes lead to the same destination: discovering and comparing products from thousands of online shops, and buying directly from the retailer – without ever having to leave the James service. 

How does an online shop become visible to AI agents?

One of the most common questions from retailers is: “What do I need to do to ensure my online shop is found by AI agents?” The answer depends on how deep the integration needs to be.

Agentic Commerce on Shopify: Discovery vs. Direct Sales

In Agentic Commerce, not all online shops are integrated into AI experiences in the same way. The e-commerce platform Shopify therefore distinguishes between two levels: simply making products discoverable within AI systems, and direct sales within an AI-supported purchasing process.

Discovery happens automatically. As soon as a shop goes live on Shopify, AI agents can access the products via the Shopify Catalogue – through scraping, feeds or the Catalogue API. The merchant does not need to take any active steps to set this up.

Native selling, on the other hand – that is, checking out directly within an AI chat (e.g. ChatGPT or Google AI) – requires the Agentic Storefronts feature to be enabled in the Shopify Admin. Without this, customers will be redirected to the online store.

Requirements for Agentic Storefronts

The following requirements currently apply to Agentic Storefronts:

• The shop must be based in the US for Microsoft Co-Pilot, Google AI Mode and Gemini (not for ChatGPT), and in all cases, sales must be made to US customers
• Acceptance of the Shopify Agentic Storefronts Supplemental Terms of Service
• Shop policies (Terms and Conditions, Returns Policy) must be in place
• Products must be eligible for the Shopify Catalogue

If these conditions are met, Agentic Storefronts will be activated automatically.

Product data in Agentic Commerce: The key to AI discoverability

For products to be found, correctly categorised and recommended in Agentic commerce systems, high-quality, structured product data is essential. Whilst Shopify supports this process using LLMs for standardisation and enrichment, the quality of the results still depends crucially on the underlying shop data.

The following best practices are particularly important to ensure that products are as easy as possible for AI agents to find and interpret.

• Search-friendly titles
  Use natural language, as customers would search (e.g. ‘men's running shoes, wide fit’ instead of internal product codes)
• Full descriptions
  Relevant information in the first 6,000 characters, including materials, warranty, compatibility and sizing
• Consistent variants
  Use consistent terminology for size and colour
• GTIN/EAN/UPC
  To ensure full UCP compatibility, each product variant should have a valid GTIN
• Real-time inventory
  Inventory data must be up to date, as outdated availability data is of no use to AI agents

Universal Commerce Protocol (UCP): The open standard for agentic commerce

For AI agents to find, compare and purchase products across different shops, a common technical language is required. This is precisely why the Universal Commerce Protocol (UCP) was developed — an open standard from Shopify and Google that standardises communication between retailers, shop platforms and AI agents.

The protocol is based on three key pillars:

1. Discovery & Negotiation:
   Both traders and agents publish profiles that outline their capabilities. Discovery is the process of retrieving these profiles, and Negotiation calculates the common intersection.
2. UCP-Manifesto:
   In much the same way as robots.txt works for search engines, an online shop hosts a ucp.json file that declares its capabilities – identity, endpoints, supported payment methods and dynamic pricing.
3. Standardised product data:
   UCP requires Schema.org markup as a baseline, extended to include state-dependent variables (e.g. ‘if purchased together with X, then price Y’).

Be present in AI-commerce even without your own Shopify shop

With the new Agentic Plan, Shopify is opening up access to Agentic Commerce to brands that do not have their own Shopify online shop. Businesses can store their product data centrally in the Shopify Catalogue, making their offerings discoverable and available via supported AI channels. Since the Winter '26 Edition, Shopify has been offering a new Agentic Plan. This allows brands that do not operate a Shopify shop to list their products in the Shopify Catalogue and make them available via AI channels. The product data is uploaded once and then distributed across all supported AI platforms.

What restrictions are currently in place (as of April 2026)?

• Local delivery and in-store collection are not yet supported in Agentic Checkouts
• Google Analytics and custom pixels do not work in all Agentic checkouts
• Agentic Storefronts are currently limited to the US market
• The endpoint URLs for the Catalog MCP may change
• Search results must not be cached – they reflect retailers’ real-time preferences

When is Agentic Commerce coming to Germany?

Shopify’s Agentic Storefronts have been in early access since January 2026 – though primarily for the US market. Eligible merchants will receive an email and a notification in the Shopify Admin when the feature becomes available for their online shop. Integrated AI partners already include ChatGPT, Google AI Mode/Gemini, Microsoft Copilot and Perplexity.

There is not yet a specific launch date for Germany and the EU. The reasons for this lie primarily in the regulatory framework:

• GDPR: Data protection requirements for AI-powered checkouts and the sharing of customer data with third-party platforms
• EU AI Act: Regulation of AI systems in the retail sector, particularly in relation to automated purchase recommendations
• Consumer protection law: Unresolved issues regarding the right of withdrawal, price labelling and information requirements for purchases arranged by agents

Whilst brands such as Poshmark and Reebok are already conducting live trials in the US, the European market is still in the observation phase.

Of interest to the European market:
With the launch of the Agentic Commerce Alliance (ACA) in May 2025, Shopware has adopted an alternative approach – focusing on open standards rather than the closed platforms of the tech giants. It remains to be seen whether Shopify’s UCP, the Agentic Commerce Protocol (ACP) or other open standards will ultimately prevail in Europe.

Recommendation for retailers:
Even if the feature isn’t available in Germany just yet, it’s worth laying the groundwork now – optimised product data, comprehensive shop guidelines, structured data (Schema.org) and real-time inventory. Those who are prepared will be the first to benefit when the feature is rolled out across the EU.

The future of shopping: How agentic commerce is transforming online retail

Agentic Commerce is still in its infancy, but it is already providing a very clear indication of the direction in which digital shopping is heading. Instead of laboriously navigating online shops, filters and comparison sites yourself, intelligent AI agents will in future take on an ever-increasing share of product search, evaluation and checkout processes.

• Personalisation:
  AI agents learn your preferences and proactively filter content
• Channel independence:
  Whether it’s video, chat, a voice assistant or a website – the agent works everywhere
• Price transparency:
  Automatic comparison across hundreds of shops
• A seamless shopping experience:
  From discovery to checkout in seconds

James is our first step in this direction. A concierge who doesn’t just search, but understands – and presents you with the best options on a silver platter.

Enquire about an Agentic Commerce project now.

Sources & further links

[1] Shopify Help Center – Requirements for Agentic Storefronts
[2] Shopify Dev Docs – About Shopify Catalog
[3] Shopify Dev Docs – Catalog MCP Server
[4] Shopify Engineering – Building the Universal Commerce Protocol
[5] Shopify News – The Agentic Commerce Platform
[6] Shopify Enterprise Blog – Agentic-Ready Product Data
[7] Google Developers Blog – Under the Hood: Universal Commerce Protocol
[8] Shopify Help Center – Agentic Storefronts
[9] Shopify News – Winter '26 Agentic Storefronts
[10] Dreizack Medien – Agentic Commerce Update 2026: Google, Shopify & JTL im Check 
[11] Cross-Border Commerce Europe – Introducing Shopify Agentic Storefronts