American Express is looking for a bigger role in retail transactions, and it wants to get involved before customers decide where to spend their money. The company’s investments in artificial intelligence, restaurant reservations and automated payments suggest an ambition to influence merchant discovery as well as the eventual purchase.
During its October 23 third-quarter earnings call, Chairman and CEO Stephen Squeri outlined initiatives involving OpenAI, personalized recommendations and technology designed to support purchases made by AI agents. The company is developing ways to connect its existing cardholder benefits and merchant relationships with AI-powered shopping experiences.
For Canadian retailers, the implications extend beyond payment processing. If consumers increasingly rely on AI assistants to find restaurants, compare offers and select merchants, the companies supplying those services could gain influence over customer acquisition. American Express has a potentially useful starting position. Its network connects cardholders and merchants, while its rewards and membership programs already provide incentives for consumers to choose particular businesses.
Restaurants Provide an Early Commercial Test
American Express’s restaurant business offers a practical example of how the company is connecting merchant discovery with membership benefits. During the third quarter, spending at restaurants increased 9%, while spending at restaurants participating in its Resy ecosystem grew 18%. Spending through Fine Hotels + Resorts and The Hotel Collection increased 50%, according to management.
Those figures demonstrate stronger reported spending growth within selected merchant programs, although they do not establish why that growth occurred. American Express has not disclosed how much of the additional spending came from new customers, existing customers spending more, or activity directly attributable to its recommendations and reservation services. The distinction matters to restaurants assessing the economics of third-party customer acquisition. A reservation that brings in someone who would otherwise have visited a competitor can create meaningful value, while a reservation from an existing customer may simply add another intermediary to the transaction.
American Express has an incentive to make these programs attractive to both merchants and cardholders. Restaurants gain access to a membership ecosystem, while cardholders receive dining benefits that encourage them to remain engaged with American Express products. The company can also participate in the customer’s decision before the bill arrives. Restaurant discovery, reservations and personalized offers provide opportunities to connect a consumer with a participating business, with American Express potentially handling the subsequent payment.
That arrangement gives the company several ways to reinforce its existing relationships. Whether it produces profitable incremental business for participating restaurants is a separate question.
Bringing Cardholder Benefits Into AI Shopping
American Express is extending elements of this approach into AI-powered consumer services. Squeri identified five areas of AI development during the earnings call, covering customer value propositions, acquisition, engagement, payments and servicing. The company has also described work with OpenAI involving restaurant discovery and reservations through ChatGPT, along with plans to make American Express Offers available through the platform.
- For consumers, the potential appeal is straightforward. An AI assistant could help identify a restaurant, find an available reservation and surface a relevant cardholder benefit without requiring the customer to navigate several separate services.
- For American Express, the commercial opportunity is to make its membership products useful during the shopping decision. A customer who encounters an attractive offer or dining benefit may have an additional reason to choose a participating merchant and pay with an American Express card.
That does not mean American Express controls which merchants ChatGPT recommends. Its integrations provide access to particular services and benefits, but the company has not demonstrated control over general AI search results or preferential treatment for merchants within its network. The commercial value of AI discovery depends partly on who controls the recommendation. American Express can develop its own merchant connections and offers, but the broader customer experience also depends on the AI platforms through which consumers access those services.
Preparing for Purchases Made by AI Agents
American Express is also investing in the infrastructure needed for AI agents to complete transactions on behalf of customers.
Its Agentic Commerce Experiences developer kit and Intent Intelligence Framework address issues such as customer authorization, purchase instructions and payment credentials. These are practical requirements for a system in which software may select products or services and initiate payment without the consumer manually completing every step.
- For merchants, authorization is particularly important. A retailer needs confidence that an automated purchase reflects the customer’s instructions, while the payment provider must be able to authenticate the transaction and manage potential disputes.
Squeri discussed the importance of approved agent access and safeguards against automated activity that could interfere with reservations or membership benefits. Those concerns illustrate the operational problems that payment networks will need to address if agentic commerce develops into a meaningful transaction channel.
The commercial opportunity is to become a trusted provider of that infrastructure. Payment networks already manage authorization, fraud controls and transaction processing, giving them relevant capabilities as AI systems begin interacting with merchants.
American Express is investing ahead of demonstrated demand, however. Management acknowledged that agentic transaction volumes remain small, and the company has not disclosed a material revenue contribution from these initiatives. Other payment networks and technology companies are pursuing their own approaches. American Express may help shape how agentic transactions operate, but the eventual standards and commercial relationships remain unsettled.
Retailers Face Another Customer-Acquisition Question
Retailers have considerable experience with intermediaries that promise access to customers. Search engines, marketplaces, delivery platforms and social media services can generate sales, but merchants must weigh those benefits against advertising costs, commissions and reduced control over customer relationships. AI shopping could add another channel to that calculation. Businesses may eventually need to make their products, availability, offers and payment capabilities accessible to systems that consumers use to research and complete purchases.
American Express could benefit by making its merchant relationships and cardholder incentives available through those systems. Its existing network may reduce some of the work required to connect participating businesses with AI-enabled services. The commercial question for merchants is whether these arrangements produce profitable sales they would not otherwise have captured. Additional transactions are useful only when the revenue remaining after acquisition, promotion, payment and fulfilment costs justifies the investment.
A restaurant receiving more reservations through an AI service, for example, would need to determine whether those bookings represent new demand or customers shifting from its existing channels. A retailer participating in an AI shopping platform faces a similar calculation when customers who previously purchased directly begin arriving through an intermediary.
There is also the possibility that merchant visibility becomes increasingly dependent on technical integrations. Businesses that are easier for AI systems to discover, reserve or transact with could have an advantage, although American Express’s announcements do not establish that such a competitive hierarchy currently exists.
For retailers, the risk is familiar even if the technology is new. A customer-acquisition channel can become expensive when merchants grow dependent on an intermediary they do not control.
The Canadian Market Has an Existing Foundation
American Express already operates premium-card, dining and merchant-benefit programs in Canada. These relationships provide a foundation for consumer and merchant services, although the company’s newer AI integrations are not all available in the Canadian market.
The Resy reservation experience described within ChatGPT has been identified as a U.S.-only offering. American Express has not confirmed a Canadian launch timetable for the broader AI-commerce capabilities discussed during its earnings call.
Canadian restaurants should therefore not interpret the company’s AI announcements as evidence that those integrations are already generating local reservations or additional sales. The more useful exercise is to consider how similar services could affect merchant discovery and acquisition costs if they become available. Restaurants, hotels and premium discretionary retailers are particularly relevant because personalized recommendations, availability and membership benefits can influence purchasing decisions. These businesses already participate in ecosystems where access to customers can depend on reservations, loyalty relationships and third-party platforms.
Canadian banks and competing payment networks also have reason to pay attention. If AI assistants become involved in selecting offers and authorizing purchases, payment products will need to function within those experiences.
That could create another area of competition beyond acceptance, rewards and card fees. Financial institutions may increasingly compete over how their products appear and operate during AI-assisted shopping, although the commercial scale of that opportunity has yet to be established.
Strong Earnings, Unproven AI Returns
American Express can fund these initiatives from an established and profitable business rather than relying on immediate returns from AI commerce. The company reported approximately US$20.1 billion in third-quarter revenue, an increase of 9%, with billed business growing 8%. Net card fees increased 17% on a foreign-exchange-adjusted basis, reflecting continued demand for products that generate recurring membership revenue.
Those results demonstrate the strength of the existing business, but they do not establish that AI is producing a meaningful financial contribution. American Express has not disclosed the volume of sales generated through AI recommendations, conversion rates associated with its integrations or the incremental merchant spending attributable to those services.
- Consumer adoption presents another uncertainty. People may be comfortable using AI to compare restaurants or identify offers while remaining reluctant to let software authorize purchases without reviewing the details.
- Merchants have their own reasons to be selective. An AI shopping service needs to deliver customers at an acceptable cost, provide reliable transaction information and avoid undermining more profitable direct relationships.
American Express is attempting to establish a role before those commercial arrangements are settled. Its cardholder relationships, merchant network and payment capabilities provide useful assets, but they do not guarantee that consumers or retailers will adopt the services at scale.
What Merchants Should Measure
Canadian retailers do not need to treat agentic commerce as an established sales channel. They do need to understand how their products, availability, offers and payment options could eventually be represented within AI shopping services.
American Express’s investments show how payment networks are preparing to participate earlier in the purchasing process. The company is developing capabilities that could connect merchant discovery, membership benefits and transaction authorization, although their financial impact remains uncertain.
For merchants, the relevant measure will be incremental profitable demand. If AI-powered services deliver new customers at a reasonable cost, they could become a useful addition to existing channels. If they merely redirect established customers through another intermediary, the commercial benefit will be considerably harder to justify.
