Autonomous agents are taking over the customer journey—from search to checkout—and companies that wait to adapt risk being cut out entirely.
The first and only buying conversation for some consumers might soon be between two digital entities: the consumer’s shopping agent and a vendor’s systems. Artificial intelligence agents—autonomous systems that act on behalf of users, from research to purchase—are rapidly reshaping how people discover, evaluate, and buy products.
In agent-to-agent commerce, features such as Amazon’s “Buy for Me” are already automating large chunks of the customer journey. Perplexity, ChatGPT, and Gemini are crawling vendor sites, reviewing options, and recommending products. In addition, payment processors and credit card companies are embedding payment services to secure their position in agentic commerce. With these fast-moving developments, many consumers are using AI to research products, and a smaller group have delegated price comparison, selection, and even checkout to agents. While few consumers let an AI agent buy items without a final signoff, the agent can still disrupt most of the journey. That final transaction layer has arrived faster than vendors expected and may soon spread to become a sizeable share of consumer purchases. What began in retail is now extending to hospitality, financial services, and other industries. In this new A2A world, companies must decide whether to lead, ride along, or get sidelined by this shift. Lessons from Food and Travel Aggregators Roughly a decade ago, aggregator firms such as DoorDash and Expedia quietly took control of the restaurant and travel marketplaces. They scraped online menus and seat inventories, built scale, then flipped the economics on suppliers. Restaurant and airline brands that waited to decide whether and how to participate with the aggregators lost leverage, forced into inferior terms if they joined the aggregators later. In hospitality, Marriott reaped substantial benefits after expanding its partnership with Expedia in 2019. Marriott already had a broad channel partnership with Expedia, but in 2019 they enhanced the partnership by adding last-minute inventory and giving Marriott greater control over the customer experience. Marriott’s Bonvoy loyalty program continues to drive traffic to the home site or through Marriott credit cards or an ecosystem of noncompetitive partners. The Expedia partnership gave Marriott better control over its customer experience, and the hotel chain’s revenues grew at a 10% compound annual rate from 2022 through 2024, compared to 1% from 2017 through 2019. The message from Marriott’s experience is clear: There can be advantages to collaborating with bots instead of just letting them market to customers in isolation. Today, agentic AI platforms are in a similar early phase—scraping listings, building recommendation layers, and enabling checkout without ever sending the buyer to a vendor’s website. For consumers, it can be easier to access one interface to shop across a wide range of channels and find the best deals. For sellers, agents make it easier to cross-list more items, in more places, more cheaply than with existing tools. Just as with the restaurant and travel aggregators, speed matters. First movers can shape the terms, while latecomers take what’s left. The Retailer’s Dilemma Product marketplaces operate along an AI agent spectrum. Consider retail, where our recent shopping experiences through AI agents found that retailers’ decisions on agents range along a spectrum of four approaches: Fully closed. In this approach, the vendor blocks AI agents from crawling its site. This works best for companies with a distinctive value proposition that keeps customers loyal even as their behavior otherwise shifts to AI platforms. The company protects its data, owns the customer relationship, and maintains its profit margins, at least in the near term. However, if AI agents become the default search option, the company could lose traffic to its own site and the ability to shape future traffic. Amazon fits the bill here. It developed its own AI agent that makes purchases on other brands’ sites. But it disallows external agents scraping its own site listings . Passive open. Listings are exposed but not optimized. The company can pick and choose which customer data, prices, and product specifications it wants to share and what it retains for its own site. On the flip side, because listings are not optimized for agent crawling, they may look less attractive than competitor offerings. That runs the risk of AI platforms recommending competitors over time if the platforms gain the upper hand at the top of the marketing and sales funnel. Pottery Barn is pursuing this approach. A subset of its products are available through Perplexity’s Buy with Pro or ChatGPT, where shoppers see listing details such as prices, color and styles, care instructions, and a synthesis of features and reviews. Shipping is free, but additional services such as a protection plan, putting items on a registry, or home design services must be done on Pottery Barn’s own site. Partial or full partnership. Some companies have actively joined a specific AI agent’s program to allows structured feeds for scraping and then a checkout experience run by the agent, rather than making the user go to the company’s own website to finish. The opportunity here lies in shaping the customer experience with the AI agent partner and, in future, possibly having priority positioning in product query results and data sharing advantages. What’s tricky for retailers is having to pick a winner—and potentially picking wrong. OpenAI recently launched Instant Checkout for US users who can buy directly in the platform from US Etsy sellers and will soon have the same option for over a million Shopify merchants. OpenAI considers the enablement of Instant Checkout along with other factors when ranking merchants that sell the same product. Fully active agent-to-agent. A vendor with sufficient market clout might decide to build its own “.bot” site that could be accessed only by AI agents. That would lay the foundation for all AI agents to access the company’s data, thereby raising the speed and efficiency of agents scraping its inventory, descriptions, reviews, prices, and other information. While several companies have discussed with us interest in this approach, none have yet committed. It would take the most effort and resources of the four strategies and runs the risk of AI platforms charging a company rents over time if the agents take control of demand traffic at the top of the sales funnel. While it’s early days for A2A commerce, one key tradeoff has become clear. AI agents can increase a vendor’s reach, but they compress profit margins and increase transparency. Adding more intermediaries to a marketplace almost always weakens the unit economics for existing vendors. Volume may grow, at times with a lower cost to acquire customers, but over the long run an intermediary will likely extract more of the value. Any vendor, therefore, faces a variation of the “prisoner’s dilemma.” If competitors open up and it does not, the vendor loses relevance. Yet if it opens up before building a strong value proposition, it risks being commoditized. An Agent-to-Agent Strategic Playbook In an A2A world, then, vendors will have to fine-tune their value proposition. They should bolster the trust and services that still make a difference in consumers’ minds. Brands will continue to stand out through their human touch, whether that consists of dedicated sales representatives, curated bundles, or post-purchase care. To thrive in this new marketplace, vendors should focus on a few dimensions: Control the checkout layer. Vendors should own the payment, shipping, and data relationship when possible, which avoids their becoming a “dumb pipe” with no way to monetize customers’ intent. Make consumers care where the transaction happens. Scarcity often creates power. Some inventory should remain exclusive to the vendor’s site, including limited products, premium bundles, or points multipliers. Create a strategic moat through value-added services available only on the company’s own site. Installation, protection plans, and service add-ons can anchor consumers to a vendor’s ecosystem, but only if it owns the last mile. Protect and monetize data. Vendors can use watermarking, tiered access to critical software, and tracking to retain visibility into consumer behavior. Play offense with generative AI engines. Feeds to agents should be optimized for visibility and inclusion in the agent recommendations. A headless .bot version would presumably have this objective. Shape early alliances. Recalling the restaurant and travel aggregator history, it pays to strike co-branded deals with agents now, while vendors still have leverage, and ensure those deals involve sharing data. It’s too soon to know how agent-to-agent competition will play out. A few mega-agents might eventually mediate most buying. Alternatively, strong specialty agents could emerge by industry, category, or lifestyle. And some company brands could deploy white-label agents to hold onto the customer relationship. Regardless of how the future unfolds, agents will reshape the marketing and sales funnel—maybe even erase it. The customer journey is no longer linear, and in some cases may not even include a vendor’s website or app. When the next consumer sends her AI agent to shop, vendors had better be ready with alternative methods of selling.
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Subscribe now: Full access to 90,000+ articles, reports, videos and imagesWe invite you to subscribe to Luxury Daily and join the ranks of the smartest luxury marketers worldwide. Our subscribers include decision-makers and executives from the world's leading luxury brands, retailers, agencies, publishers, market researchers, universities and consultancies.
Read more »
What Should Retailers Do About AI Shoppers?Autonomous agents are taking over the customer journey—from search to checkout—and companies that wait to adapt risk being cut out entirely.
Read more »
What Should Retailers Do About AI Shoppers?Autonomous agents are taking over the customer journey—from search to checkout—and companies that wait to adapt risk being cut out entirely.
Read more »
Brooks Nader styles a sheer corset with baggy jeans at Celsius Spritz Vibe launch in MiamiThe star-studded, aprés ski-themed party celebrated the brand's first-ever limited-edition flavor: Spritz Vibe, an icy, velvety citrus mix. (It's available through December at major retailers, including Amazon, while supplies last.)
Read more »
DoorDash Launches Emergency Food Response Amid Government Shutdown, Offering Free Delivery to SNAP RecipientsDoorDash is providing free delivery and waived fees for SNAP recipients as federal assistance faces disruption, partnering with food banks and retailers to deliver food to those in need amid the ongoing government shutdown. The program aims to deliver an estimated 1 million meals next month.
Read more »
Why ESIM Is An Underrated Technology That Businesses Shouldn’t IgnoreLow awareness is stalling adoption of eSim technology. Here's how travel brands, retailers and mobile operators can close the gap and unlock new revenue through smarter, more connected experiences.
Read more »