“Agentic commerce” has driven retail industry discourse over the past year, but within the past six months, the drumbeat of new launches, partnerships, and integrations has intensified. The pace really picked up when Shopify staked its claim in the “agentic era,” partnering with OpenAI to enable checkout within ChatGPT for its community of sellers, which range from Glossier to David’s Bridal and, most recently, Estée Lauder; not to mention the vast community of small businesses that rely on the platform to power their commerce businesses. Industry giants like Walmart and Target soon followed, announcing plans to integrate their product feeds and unique services into the platform.

In the last few weeks alone, even more companies have joined the new race, adding new features and capabilities that support shoppers as they browse and buy. Behind the headlines and hype, most “agents” are doing less of the transactional heavy lifting and are acting more like experimental co-shoppers. Yet that doesn’t negate the new reality: just as consumers flocked to Google, they are now turning to generative AI platforms like ChatGPT, Perplexity, Claude, and Gemini for information, answers, and insights.

The main difference, though, is that consumers no longer need to sift through pages of information to find what they need. Through a conversational exchange between human and bot, they will not only get relevant information but also an in-depth experience that makes finding inspiration, comparing products, and finding the best deals easier. 

Are we in the moment of reckoning that many are making it out to be? With this Member Brief, we examine: 

• The latest developments in AI-powered shopping experiences and agentic commerce.
• How conflicting definitions of “agentic commerce” are actually making us lose sight of how (and why) AI is truly transforming commerce.
• What merchants can do today and in the near future to ready their eCommerce experiences for agents.
• Why now is the time for merchants to say goodbye to the tired eCommerce playbook.

‍

The Agentic Race is at the Top of the Funnel

Google launched what Matt Maher of M7 Innovations calls the company’s “most aggressive AI shopping tools yet.” Many deemed the tech giant “doomed” early in the AI hype cycle (we disagreed). Still, these new capabilities allow Google to take the heavy mental and physical load of shopping off consumers: AI can call stores to check stock levels and gather sale details, can match shopping needs and requests against more than 50 billion product listings, and yes, even autonomously purchase items via Google Pay. Agentic checkout is arguably the most compelling value proposition, mainly because Google will track an item based on specific size, color, and desired price. Consumers will receive a notification when all parameters are met and, if the merchant is eligible, Google can buy it on the merchant’s site using Google Pay. 

More recently, Perplexity announced it will roll out free agentic shopping, developed with PayPal, which will allow “seamless purchase right from the answer,” according to Perplexity Chief Business Officer Dmitry Shevelenko. These new features suggest a near future in which agentic browsers sift through inventory, hunt for deals, and build carts for us as we go about our days. They would make decisions and make purchases without our intervention. 

Some analysts note that we’re very far off from that reality. In practice, the user experience still requires many (many) steps that require human approvals, substitutions, and corrections before anything gets to a doorstep. And in the case of Perplexity, while the tool is described as “agentic,” it appears to enable purchases on the platform, with PayPal merchants handling transactions, customer service, and returns. The AI bot isn’t taking over the purchase experience for the customer. The “agent” may remove a few steps from the linear customer experience, but the friction points feel so familiar. 

“The point of agentic commerce is that it’s in a decision-making role, and it can execute on those decisions. What it actually does is invert the paradigm of where the human sits in the loop,” explained Heather Hershey, Research Director at IDC, who leads global digital commerce strategy. “To determine whether or not a product is actually agentic, it's whether or not it's responding to a human operational prompt. Because it could be computing things agentically in the background—you can see this in Perplexity and ChatGPT—they show you their homework while it's doing it. But just because that platform is an agent, it doesn't mean that all the activities going on in that platform are agentic.”

However, some call this a game of semantics. Mike Mallazo, director of Honey at PayPal and Future Commerce contributor, noted that “agentic” is being used as a catch-all to conflate different innovations enabled by AI. One, of course, is the earlier example: when agents autonomously complete a purchase on a user's behalf. The other is when LLMs provide relevant responses to more complex, conversational queries. Both scenarios point to staggering new realities of how consumers participate in commerce. In Future Commerce’s survey of 1,000 consumers, respondents said they used AI platforms to support basic brand research (38%), product and category research (36%), to compare specific products from different brands (29%), to get ideas and inspiration for a particular need (29%), to research a trend (25%), and to research the best product in a specific category (24%). These behaviors account for most of the shopping journey, underscoring the widespread use and influence of AI platforms. This influence is poised to grow: nearly half of consumers say they're likely to turn to generative AI platforms first for gift shopping ideas and recommendations this holiday season (49%).

OpenAI realizes the TOFU and MOFU power of its tool, which is why it has introduced an entirely new shopping experience focused on aiding discovery, product comparison, gift selection, deal hunting, and even finding lookalike products.

An expanded ecosystem of AI features that spans the most influential stages of the shopping journey raises questions about merchants’ roles in the journey. There will always be a need for products and, in turn, brands.

But what role will the branded environment, the eCommerce site, play in the future? What is the purpose of the eCommerce site when consumers are going to a “neutral playing ground” to get everything they need? 

Agentic commerce today is less about “AI doing all the shopping for you” and more about “AI changing where and how customers decide what to buy.” It compares items, narrows brands, and validates decisions. It can even find a timely promo code. When nearly half of consumers maintain a perpetual shopping list, an AI agent can help reduce the mental and emotional load. But the reality is that it can’t account for the emotional and psychological intricacies that drive decision-making, from brand preferences to product selection. 

With all of this nuance in mind, we believe eCommerce sites today will be designed with two very distinct roles and goals in mind: 

1. To be a source of absolute truth, where rich product data, social proof, and the latest deals and policy information live.
   
   
2. To be a destination for immersive brand storytelling and experience, which even the most sophisticated generative AI platforms can’t replicate.
   

The merchants who succeed in the “agentic era” will be those who deliberately design both.

A Brave New World of Search and Discovery (But not Yet Seamless Shopping)

It’s easy to understand why retailers are so eager to hop on the agentic bandwagon, or at least start building a plan. Our New Modes research found that most consumers are in a perpetual state of purchase consideration. They’re always thinking about what they need (or want) to buy, and platforms like ChatGPT, Perplexity, Gemini, and Claude turn these sometimes-complex inquiries into logical and streamlined outputs. 

Gen Z and Millennials are leading this movement, with 43% using these tools daily, largely because generative AI platforms are completely reframing the top and middle stages of the funnel, when we have our most abstract thoughts: “How do I dress Quiet Luxury?” “What should I buy for my 8-year-old nephew who loves comics but has everything?” 

When we use a traditional search engine, we typically write a keyword or phrase to “game” the engine and get the best, most accurate answer for our needs. Over the years, that process has become more semantic, with Google’s BERT language model and Neural Matching, according to Tea Benes, SEO Strategist at IM Digital. ChatGPT, Gemini, Claude, and Perplexity have entirely different end goals, focusing less on retrieving and synthesizing information and more on generating answers. For consumers, that means every prompt and question becomes an output that’s hyper-tailored to their needs. 

“Traditional search excels at finding and ranking the single best document to answer your query,” Benes said. “The new AI-driven platforms excel at extracting facts from multiple sources to construct a direct answer. This changes everything.” By performing a “query fan-out,” these platforms can break complex questions into “smaller, factual sub-questions,” Benes explained. In the case of Google’s AI Overviews, “it might send one part of your query to its Shopping Graph to get the latest price, while another part goes to the web index to find product specifications. It's a way of gathering evidence from the best possible source for each piece of the puzzle before synthesizing an answer.” 

Once the “work” is done on the back end, these systems go beyond semantic search, which focuses on providing the most topically relevant information, and ask follow-up questions to confirm specific details and construct a direct answer. It’s why, when we sometimes think we’re asking ChatGPT for straightforward, specific information, the bot responds with a list of five follow-up questions. 

Benes provided an example: “When it sees a query like 'a lightweight laptop under $1,000 with more than 8 hours of battery life,' it breaks it down and starts looking for specific data points. It asks questions like: 'What is this laptop's weight? What is its price? What is its official battery rating?'”

 These differences are what make these platforms so disruptive. Consumers no longer have to hunt for information; they don’t have to “game” a system to get measly crumbs of data, hoping that they’re on the path to an accurate answer. It is an intimate, lightning-fast exchange of information, with each source (man and machine) building and iterating together. Rich, contextual interactions like these require a completely different approach to content creation and, in turn, eCommerce experience design.

For instance, a retailer selling high-performance gear could SEO-optimize the product page for a specific jacket so it ranks perfectly for “men’s rain jacket,” Benes explained. With great images, compelling copy, and a solid backlink profile, it’s in good shape. But with an agentic query, the basics fall short if core information, such as fabric type, waterproof rating, and other factors, is unstructured or buried in different places, like PDFs. “The well-ranking page is a data black box.”

This is the heart of the AI challenge. Generative AI platforms and agents don’t care how pretty your page design is. They just care that your data is rich, legible, and accessible. 

Ananda Chakravarty, VP of Research for IDC Retail Insights, noted that although the “agentic promise” has not been fully realized, “there’s a lot of fear among retailers that they’re going to eat into their eCommerce business,” he said. “It’s all a wait and see; they don’t have the technology to match.” 

Agents are astonishingly strong at discovery and comparison, and it’s changing the conversion math. From Nov. 1 to Dec. 1, 2025, Adobe reported that AI traffic to US retail sites (measured by shoppers clicking referral links) increased by 760%. Our Q4 2025 research confirms the click-through opportunity, as 77% of consumers say they would visit a brand or retailer's eCommerce site after receiving an AI recommendation, rather than purchasing directly in-platform. These visitors arrive with compressed research cycles and validated intent, so the site experience must be meaningfully different and serve that customer differently.

Answer engines determine which sites to include in the new consideration set, so the key is to ensure your product and brand are part of the AI mix, and that your eCommerce experience is delightful and memorable for humans once they arrive.

A Destination Centered on Data Transparency and Truth

Merchants selling highly technical, functional products can use their eCommerce sites as a rich source of truth. AI agents trust these sources, Benes notes, because the data is rich, integrated, and up-to-date. “The site’s value won’t just be about direct sales, but how often its data is used to build AI answers.”

If a brand is integrated into generative AI responses, it is part of the contextual commerce conversation, creating a clearer path to conversion as inspiration becomes intent. In the case of ChatGPT’s shopping tool, it searches the web for price, availability, reviews, specifications, and images, then returns only the most relevant options. Brands can’t be a part of this mix if any of these data sets are incomplete, inaccurate, or inaccessible.

To achieve this, merchants must be “data-first,” not “website-first,” according to Benes. Data should be managed independently of how it's displayed. “This shift to AI creates a new sense of purpose around fundamentals, like writing detailed product descriptions that answer pre-purchase questions, or populating granular attributes like dimensions, materials, and compatibility,” she said. 

“This is the exact data that helps customers make a purchase decision. It really proves the point that you can't build a strong AI presence on a weak content foundation. The source of truth has to be rich with detail.”

Wondering how to keep up? Here are some tactical actions merchants should take: 

• Enrich Product Data: Treat key attributes as data inputs. House them in the PIM or eCommerce backend.
  
  
• Surface Data with Schema: Implement a detailed product schema to clearly define all key product features as distinct, machine-readable attributes.
  
  
• Optimize the Product Feed(s): Add specific attributes to your product feed, starting with Google Merchant Center. This gives AI systems a direct line to your official product data and future-proofs your brand for Google and other AI platforms that want a structured data feed.
  
  

Generative AI platforms are simply another destination for consumers to engage with commerce. Just as marketplaces, social platforms, and media sites do, ChatGPT, Perplexity, and other AI platforms seek clean, consistent, and accessible data. That data is the link between your brand and consumers who may not even know it exists. Data helps you become a “trusted citation,” a building block for creating the rich, accurate answer consumers now expect from these platforms. Benes added that treating your business like a data platform, and “making your product catalog and checkout logic as clean, consistent, and accessible as possible,” allows a future AI partner to “easily plug into it.” 

A Space for Curation and Rich Storytelling 

When consumers have access to all the information they could want or need elsewhere, the eCommerce site can play a new role in their lives: curator, community-builder, and entertainer. 

58% of global consumers agree: “It bothers me when a brand I love has a poor website experience. 

This is where merchants have the unique opportunity to embrace lore. The stories that built their brand, their products, and their history are clear differentiators that “answer synthesizers” simply cannot replicate. Interactive tools such as product customization and virtual try-on add context to considered products and validate purchase decisions initiated on an AI platform. Video storytelling, curated merchandising, and trend-based editorial create a sense of worldbuilding around a brand, elevating its unique point of view and sense of taste, a vast departure from AI platforms, which are powered mainly by logic and tend to validate the end user’s thoughts and opinions. 

While AI platforms provide ease and utility, beauty and engagement can still live on the eCommerce site. For many, this is where brand loyalty is found and nurtured, especially when brands integrate community-building spaces and data-driven personalization. But achieving this level of connection when discovery, product research, and even product comparison increasingly occur in different places requires merchants to abandon the “best practices” of eCommerce site design that have created an undeniable sea of sameness. 

“It stops being just a digital store and becomes more of a digital flagship,” said Benes. “The ultimate source of facts for machines, and the ultimate brand experience for people."

Where Man and Machine Converge

Agents can compress facts and synthesize information, but they cannot live your origin story, feel your fabrics, or remember what it was like the first time a customer walked into your store. All of those things remain within your owned experiences, and only your brand can communicate them through visual, verbal, and written storytelling.

Of course, there is a world where the two lanes, data richness and immersive experience, can coexist. When brands favor imagery with high editorial value over text, they should separate the presentation layer from the data layer. “For the human shopper, the design can stay clean and visually focused,” Benes said. “All the rich, factual data that an AI needs, like dimensions, materials, or compatibility, is still placed on the page, but it's neatly organized inside common UI elements like tabs, accordions, or 'read more' links.” 

This creates balance. A world where the logic and creativity of commerce, or the head and the heart, can coexist. It also creates a hedge against the “shitty robot future” we keep being promised. If agents will always need human correction and context, then brands should design their eCommerce sites as places people actually want to visit, not just endpoints for a machine-built cart. These branded eCommerce experiences should be sources of truth and stages for storytelling that resonates.

‍