When we talk about AI shopping agents, the conversation often starts with convenience. An assistant that can find a product, compare alternatives, and complete a purchase sounds like another way to make online shopping easier. But I’ve been wondering what happens when that assistant starts moving through an ecommerce website on the customer’s behalf. Who decides what it can access? And who is responsible for the experience it creates? Recent developments involving Amazon, Meta, Shopify and American Express suggest these questions are becoming more than theoretical.

Two different approaches to AI-assisted shopping

In September 2026, Amazon restricted Meta’s Muse AI assistant from shopping on its marketplace. According to GeekWire’s reporting, Amazon said Meta had not obtained permission for the agent to access its website and raised concerns about how it identified itself and handled customer information. On the other hand, Meta said that Muse uses secure systems to handle credentials and seeks customer approval before sensitive actions. The disagreement reflects different views about how third-party AI agents should interact with ecommerce platforms.

Shopify has taken a different approach, supporting AI-assisted product discovery and checkout through its agentic-commerce infrastructure. On 14 October 2026, the company expanded access to its product catalogue, allowing a wider range of AI agents to discover products and facilitate purchases without requiring individual partnerships. Customers must still approve purchases, while Shopify provides the infrastructure connecting agents, merchants and checkout. Shopify has also introduced support for browser-based AI agents to complete purchases on participating merchants’ websites with customer authorization.

This difference is important. Amazon is restricting unauthorized third-party access to its marketplace, while Shopify is developing structured, supported ways for AI agents to interact with merchants. Neither approach necessarily means unrestricted access or complete opposition to AI-assisted shopping. The disagreement is more about the conditions under which agents can participate.

Who controls the shopping experience?

Traditionally, retailers have had considerable control over what customers see on their websites. They decide how products are presented, how categories are organized, which information appears during checkout and how offers are communicated. An AI shopping agent could change that relationship.

Instead of browsing a website, a customer might ask an assistant to find a suitable product within a budget, compare delivery options and recommend where to buy it. The retailer may still provide the product, fulfil the order and handle customer service. But the agent could influence which products are considered and how their information is presented.

That creates opportunities for easier shopping, alongside questions about accuracy, transparency and customer expectations. The question of control also extends to the transaction itself. On 13 October 2026, American Express introduced its Intent Intelligence Framework, designed to help verify that purchases initiated by AI agents match the customer’s original instructions. The proposed framework addresses a practical concern: how can payment providers establish that an agent is acting within the limits of what the customer authorized? It is an emerging approach rather than a universally implemented solution. But it shows us how the conversation is expanding beyond access to retailer websites towards authorization, accountability and transaction security.

What this means for ecommerce businesses

For now, there is no single approach to agentic commerce. Retailers, marketplaces, payment providers and technology companies are making different decisions about access and control. However, the implications for ecommerce messaging are already worth considering.

If AI agents increasingly interpret product information on behalf of customers, businesses will need to pay attention to how clearly they describe products, conditions and customer promises. Information that is understandable to a person browsing a website also needs to remain accurate when another system retrieves, compares or summarizes it. That includes product attributes, availability, prices, delivery conditions and other details that agents may use when presenting options to customers.

For businesses operating across several markets, the challenge becomes even more complex. Product information, delivery promises and purchasing conditions may differ from one country to another, even when the products themselves are identical. An agent that presents outdated or inconsistent information could create expectations the retailer cannot fulfil. And when the customer reaches checkout or needs support, the information presented by the agent should not conflict with what the retailer can actually deliver. This is where the discussion moves beyond technology.

As AI agents take on more shopping tasks, retailers will need to consider not only how they present information to customers, but also how that information is interpreted and communicated by other systems. An agent may navigate a retailer’s website or interact with its commerce infrastructure, but the customer may never see the original product page or the context in which the information was presented. For ecommerce businesses, the challenge is to make sure that product information, purchasing conditions and customer promises remain clear, accurate and consistent, regardless of how customers choose to shop. The shopping experience may change, but the importance of clear communication does not.

Sources

GeekWire: Amazon blocks Meta’s Muse AI assistant, 20 September 2026.

Practical Ecommerce: Shopify Opens to Meta Muse; Amazon Balks, 24 September 2026.

The Verge: Shopify opens its catalog up to more AI agents, 14 October 2026.

TechCrunch: Shopify opens checkout to browser-based AI agents, 28 September 2026.

American Express: Intent Intelligence Framework for Agentic Commerce, 13 October 2026.

———————————————

Written by Iusti Ikert, Founder of In Between the Lines — a messaging and customer experience alignment studio for scaling product-led brands. I work with growing DTC companies to align brand promise, marketing, and operational delivery so they can scale without friction.

Connect on LinkedIn or follow on Instagram for insights on sustainable growth, messaging clarity, and customer experience alignment.