Agentic Commerce: Is the UK Ready for Autonomous Shopping?

In recent months, there has been considerable buzz surrounding agentic commerce, touted as the next significant advancement for online retailers. Many experts warn that businesses should prepare their websites for this new wave of technology or risk being left behind. While some assert that artificial intelligence (AI) is ushering in the most radical transformation in e-commerce since the smartphone’s inception, I believe we are still quite far from agentic commerce becoming a mainstream method of shopping in the UK, let alone fully autonomous.

What is Agentic Commerce?

Agentic commerce refers to the use of AI systems designed to perform tasks autonomously, rather than merely providing information. To illustrate, think of generative AI, like ChatGPT, as a recipe book that offers information, while agentic AI is akin to a personal chef that not only gathers ingredients but also prepares the meal. In the context of e-commerce, an agent possesses the agency to navigate websites, utilize tools, and complete transactions independently to achieve a specified goal set by the user.

How Do Agentic AI Agents Shop?

Agentic commerce operates on several levels, ranging from AI agents that browse the internet to compile products for user review to fully automated purchasing processes that require minimal human interaction. For instance, AI agents can negotiate with one another throughout the entire shopping journey, from product discovery to fulfillment.

A recent report by McKinsey illustrates this automation curve, outlining various levels of agentic commerce and their applications in e-commerce. Understanding these levels is crucial for businesses looking to adapt to this emerging landscape.

The Shopping Process with Agentic AI

When an AI agent begins the shopping process, it starts with goal initialization. For example, a user might request, “Find me a waterproof hiking jacket for under £200 suitable for a trip to the Peak District that fits in my backpack and can be delivered by Thursday.” The agent then enters the discovery phase, querying product manifests and comparing specifications, reviews, and delivery options across multiple retailers.

In the execution phase, the agent does not merely present results; it interacts with websites directly, selects the appropriate size, and adds the item to the cart. Finally, it utilizes secure payment tokens to finalize the transaction, logging the experience to improve future suggestions.

Barriers to Adoption

For agentic commerce to become fully automated, significant shifts in user behavior are necessary. Users must establish delegated authority by setting up an AI persona with specific brand exclusions and budget limits. Additionally, linking a digital wallet to the AI and implementing strict spending rules is essential to prevent unauthorized purchases.

The Protocols of Agentic Commerce

For agentic commerce to function effectively, AI must communicate with retailers and payment providers using standardized protocols. Several key protocols are currently being developed:

• Universal Commerce Protocol (UCP): Co-developed by Google and major brands like Shopify and Etsy, UCP is an open-source standard that covers the entire shopping lifecycle, allowing AI agents to understand merchant checkout logic.
• Agentic Commerce Protocol (ACP): Launched by OpenAI and Stripe, ACP focuses on enabling “Instant Checkout” within interfaces like ChatGPT, using a Shared Payment Token (SPT) for secure transactions.
• Model Context Protocol (MCP): Introduced by Anthropic, MCP serves as an open standard for AI agents, allowing them to access real-time data from retailers’ inventories and shipping databases.
• Agent Payments Protocol (AP2): This protocol ensures secure payments and provides proof of consent, meeting regulatory requirements in the UK and EU.

Preparing Your E-Commerce Site for Agentic Commerce

To be “agent-ready,” UK businesses must take several steps to prepare their e-commerce platforms:

1. Ensure Your Products Can Be Found

Prioritize implementing product Schema.org markup using JSON-LD to provide essential context regarding price, availability, and return policies. Standard identifiers like GTINs and SKUs are critical, as agents rely on these for data cross-referencing.

2. Simplify Design and Functionality

Streamlining your site’s design and functionality may require disabling features that have previously been effective. Agents need “stable selectors,” meaning that if your “Add to Basket” button ID changes during A/B testing, the agent may fail to complete the purchase.

3. Create a Path to Purchase for AI Agents

Consider developing a dedicated infrastructure for AI agents, known as a bot-friendly fast-track lane. This lane would provide a streamlined, data-only checkout flow designed specifically for authorized AI agents, minimizing the need for visual navigation intended for human users.

Frequently Asked Questions

Note: The landscape of agentic commerce is evolving, and businesses should stay informed about technological advancements and consumer behavior changes to adapt effectively.