Agentic commerce: How AI is becoming retail’s new front door

AI assistants are beginning to influence what consumers discover, compare and buy. As agentic commerce develops, retailers must make their products visible to AI without losing control of the customer relationship.

AI is becoming a new gateway to retail, changing where consumers discover products, compare options and, increasingly, make purchasing decisions.

Instead of moving between search engines, marketplaces and retailer websites, shoppers can describe what they want to an AI assistant and receive recommendations based on their needs, budget and preferences. In some cases, they can move from that conversation towards checkout without leaving the AI platform.

This is the emerging world of agentic commerce: AI systems that can understand a shopper’s intent, research options, make recommendations and, with the customer’s permission, take actions on their behalf.

Fully autonomous shopping remains at an early stage. But AI-mediated product discovery and comparison are already changing the route consumers can take towards a purchase.

McKinsey’s 2026 research found that 38% of surveyed consumers in France, Germany and the UK were using AI to research products or inform purchase decisions. Comparing brands, models, prices and reviews was the most common activity, followed by product discovery and learning about categories.

For retailers, the significance is clear: AI is moving upstream in the customer journey. It is no longer simply a tool retailers can use to recommend products. It can increasingly become the interface between the shopper and the retailer.

AI is moving upstream in the customer journey

Online retail has traditionally been built around the assumption that consumers will navigate much of the shopping journey themselves.

A customer might begin with a search engine or social platform, visit several retailer and marketplace websites, compare products and reviews, and eventually decide where to buy. Retailers have invested heavily in search engine optimisation, paid search, marketplaces and social media to attract customers to their digital storefronts.

AI introduces another layer.

A shopper looking for noise-cancelling headphones under £200, for example, can ask an AI assistant to identify suitable products, explain the differences and recommend options based on their priorities.

OpenAI’s shopping research feature illustrates the model. It allows users to describe what they are looking for and generates a personalised buyer’s guide using information such as price, availability, reviews, specifications and images. Users can refine the results as the research progresses.

The result is a change in where product consideration takes place. A retailer’s website may no longer be the first place where a potential customer encounters or evaluates its products.

That does not mean conventional search is disappearing, or that consumers will stop visiting retailer websites. AI is creating another route into retail, particularly during discovery, research and comparison.

The distinction between AI-assisted discovery and autonomous purchasing is important. McKinsey’s research shows that European consumers are using AI most strongly upstream of the transaction, where they research products, form preferences and narrow their options.

For retailers, the immediate challenge is therefore less about preparing for AI agents that automatically buy everything and more about ensuring their products are visible and accurately represented when consumers ask AI what they should buy.

AI platforms are becoming shopping channels

Some of the clearest signs of this transition can be seen in conversational shopping experiences developed by major technology and retail platforms.

Amazon has progressively turned product search into a conversational experience. On 13 May 2026, it renamed its Rufus shopping assistant Alexa for Shopping, bringing its shopping capabilities into the wider Alexa experience.

Customers can ask shopping questions using everyday language, receive product and category information, compare products and use AI-assisted features to identify deals and build baskets. Amazon says Rufus helped more than 300 million customers research, compare and buy products in 2025. Alexa for Shopping also offers features including personalised shopping guides, dynamic product comparisons, price history and automated purchasing.

The significance goes beyond Amazon’s own marketplace. For decades, retailers have designed digital storefronts around categories, search boxes, filters and product pages. Conversational AI changes that interface.

Instead of requiring customers to understand how a retailer’s catalogue is organised, the system attempts to understand what the customer needs.

Similar changes are taking place outside traditional retail marketplaces.

OpenAI launched shopping research in ChatGPT in November 2025, allowing users to research products, compare options and create personalised buying guides. In March 2026, it expanded product discovery with richer visual browsing and side-by-side comparisons.

OpenAI has also developed the Agentic Commerce Protocol, which allows merchants to provide product feeds and promotions for product discovery in ChatGPT. The company says retailers including Target, Sephora, Nordstrom, Lowe’s, Best Buy, The Home Depot and Wayfair have integrated with the protocol for discovery.

The technology is also moving closer to the transaction.

JD Sports, for example, announced plans to allow US customers to search for and purchase its products through AI platforms including Microsoft Copilot, Google Gemini and ChatGPT. The retailer is working with technology partners to connect AI-driven product discovery with checkout and payments.

The implication is that AI shopping will not be confined to retailer-owned assistants.

General-purpose AI platforms can become shopping channels in their own right. A consumer may begin a purchase journey with a question rather than a retailer, marketplace or even a clearly defined product category.

Product data is becoming a retail asset

This creates a new distribution challenge.

If AI systems are going to identify, compare and recommend products, they need reliable information about what retailers sell.

Product descriptions, specifications, prices, availability, inventory, delivery options and other attributes can become inputs into an AI-mediated shopping journey. If those inputs are incomplete, outdated or inconsistent, a retailer risks having its products poorly represented or overlooked.

High-quality product data is therefore becoming more than a merchandising or e-commerce concern. It is increasingly an asset that can determine how effectively products are discovered across digital channels.

OpenAI’s shopping systems, for example, can use merchant product data supplied through its Agentic Commerce Protocol, alongside publicly available product information and other retail sources. Its product discovery system is designed to provide more complete and up-to-date information to shoppers.

For retailers, the practical lesson is straightforward: product information needs to work beyond the product page. It may be read and interpreted by external AI systems that consumers use to make purchasing decisions.

This has helped fuel interest in generative engine optimisation, or GEO. The term broadly describes efforts to improve how brands, companies and products are represented and surfaced by generative AI systems.

GEO should not be treated simply as a replacement for SEO. Search engines will remain important, as will marketplaces, social platforms and retailer websites.

The more immediate priority is ensuring that product information is structured, accessible, consistent and current wherever AI systems obtain information used to make recommendations.

Retailers may also need to monitor a new form of digital visibility: not simply where they rank in search results, but whether their products appear when consumers ask AI assistants relevant shopping questions and whether the information presented is accurate.

Who owns the customer relationship?

Greater exposure through AI platforms comes with a strategic trade-off.

A retailer may gain access to consumers through an AI assistant while simultaneously surrendering some control over the interface through which the purchase decision is made.

That raises a fundamental question: who owns the customer relationship when an AI system makes the recommendation?

Retailers spend years building brands, loyalty programmes, digital experiences and customer relationships. If shoppers increasingly delegate product research and comparison to AI assistants, retailers may have less influence over the moment when customers decide what to buy.

McKinsey has warned that retailers that fail to become “agent ready” could lose visibility and direct customer relationships as third-party AI platforms increasingly mediate commerce. In a more extreme scenario, retailers could find themselves competing largely through factors such as price, availability, delivery speed and returns rather than through the broader brand experience.

That outcome is not inevitable.

OpenAI’s approach, for example, is designed to allow merchants to remain the merchant of record, retaining responsibility for fulfilment, returns, customer support and communication even when the customer begins the journey in ChatGPT.

Retailers can also build their own conversational shopping experiences while making their catalogues available to external AI platforms. Websites, apps, stores, loyalty programmes, customer service and fulfilment can continue to provide opportunities to establish direct relationships with customers.

The strategic challenge is therefore not simply whether to participate in AI commerce. It is how to benefit from the reach and convenience of AI platforms without allowing the retailer’s brand and customer relationship to disappear behind them.

Trust will shape autonomous shopping

The strongest argument for agentic commerce is convenience. One of its biggest constraints is trust.

Consumers may be comfortable asking an AI assistant to compare televisions, recommend trainers or identify a laptop within a particular budget. Giving the same system permission to spend money without reviewing the final choice requires a different level of confidence.

Trust has several dimensions.

Consumers need to trust the recommendation itself. They may want to understand why a particular product has been suggested and whether commercial relationships or promotions have influenced the result.

They must also trust the transaction. Delegating a purchase requires confidence that the AI has selected the correct product, paid an acceptable price and used the appropriate delivery and payment options.

There is then the question of what happens after the purchase. Returns, refunds, warranties, delivery problems and customer service become more important when the shopping journey begins through an intermediary rather than directly with the retailer.

These considerations suggest that agentic commerce will develop at different speeds across retail categories.

Consumers may be willing to automate repeat purchases of familiar household essentials long before they are comfortable delegating the purchase of a car, expensive electrical product or luxury item.

Retailers therefore need to prepare for varying levels of AI involvement rather than assume that fully autonomous purchasing will become universal.

What retailers should do now

Agentic commerce remains an emerging model, but the direction of travel is becoming clearer.

McKinsey estimates that agentic commerce could mediate between $3tn and $5tn of global consumer commerce by 2030. That is a forecast rather than a guaranteed outcome, but it indicates the potential scale of the shift.

Retail leaders do not need to predict exactly when autonomous shopping will become mainstream. They need to prepare for an AI-mediated retail environment that is already beginning to emerge.

Four priorities stand out.

Audit product data. Retailers should check that product descriptions, specifications, prices, stock, delivery information and other key attributes are accurate and consistent across relevant channels.

Monitor AI visibility. Retailers should regularly test major AI assistants with relevant shopping queries to see whether their products appear, how they are described and whether important information is correct.

Prepare for AI integration. E-commerce and technology teams should understand the product feeds, APIs and emerging standards that can connect retail catalogues with AI shopping services.

Protect the direct relationship. Retailers should decide which parts of the customer journey they want to keep within their own websites, apps, loyalty programmes and service channels, even as they use third-party AI platforms to reach new customers.

Physical stores, retailer websites, search engines, marketplaces and social media will all remain important.

But AI is creating another front door to retail.

The retailers best placed to benefit will be those that make their products easy for AI systems to discover and understand while giving consumers compelling reasons to maintain a direct relationship with the retailer behind them.