AI Shopping Agents & Conversational Commerce: FOMO or a New Frontier for Growth?
Looking to drive innovation within your enterprise by launching AI shopping agents, but don’t know where to start?
Watch our recent interview featuring Brian Gilmore, EPAM’s Global Head of Commerce GTM, and Jonathan Rioux, our Managing Principal in Artificial Intelligence, who sat down to chat at Shoptalk Spring 2025 in Las Vegas.
In this insightful interview, they dive into two of the hottest topics dominating conversation: AI and commerce. The discussion unpacks how AI is revolutionizing the marketplace, with innovations like AI shopping agents and conversational commerce transforming the way transactions happen.
Key questions addressed include:
- How much disruption is AI bringing to the retail industry?
- What opportunities and challenges does this trend pose for retailers?
AI Shopping Agents & Conversational Commerce
FOMO or a New Frontier for Growth?
AI shopping agents. For some retailers, this is an intriguing, and occasionally troubling, topic. Opinions are largely driven by organizational size and where a company resides on the maturity curve. According to our just-released AI report, 14% of execs and engineers see themselves as beginners, 32% identify as competent and 49% say they are advanced. To help retailers, all of them, get a purchase on AI agents, Brian Gilmore, EPAM’s Global Head of Commerce GTM, and Jonathan Rioux, our Managing Principal in Artificial Intelligence, explored six key topics.
1. Beyond the Stars: Experimentation with Product Reviews
So how are clients and customers currently experimenting with AI?
Gilmore notices retailers wrapping an AI agent around their product reviews so that shoppers can ask: “What did everybody say about this product?”
And is review-wrapping a good place to start or merely an instance of FOMO-ism?
“I would say it's useful,” says Rioux, cautioning that such review-based experiments must be “contextualized to how high quality your reviews are.” If you’re doing this, you must consider what discrete information you can get out of it.
Retail starring is a major element of retail culture, but it’s limited and often extreme. (“It's one or five all the time,” says Gilmore.) Is a one-star always a one-star review? Do five stars always deserve the full five? “When you look at the actual nuance, you're able to uncover some insights that you don't get with that arbitrary system,” Rioux says. You want to seek out those reviews that align with one’s tastes and worldview and those that say “I love this product but…”
It’s also an easily understood data domain. “There's not a whole lot of dependency,” Rioux adds. “You don't have to re-platform your entire company or… involve half the company to be able to derive some insights.”
2. Composable AI: “You Don’t Have to Get Ready if You Stay Ready”
Our conversationalists say that composable commerce and AI have much in common. So much, in fact, that they can make a strong, effective team.
“The same way that you’re building loosely coupled, heavily specialized, easily integratable systems together,” says Rioux. “Well, you can do this with AI.”
Illustrating the point, Rioux notes that he often asks customers to think in terms of basic questions about what they’re trying to do —
- Are you trying to summarize?
- Are you trying to generate?
- Are you trying to translate?
- Are you trying to correct?
— and then assemble those into a system that will allow them to do what they want.
This building-block approach enables astonishing retail flexibility. “When we talk about composable, the very nature of this is, at the heart of it, it's the flexibility,” says Gilmore.
Rioux agrees and says, “The name of the game in AI is speed,” and “if you can learn faster, if you can act faster, if you can react faster than your competition, this is how you win.”
Simply put: Composable AI keeps retailers prepared. As Rioux says: “You don't have to get ready if you stay ready.”
3. Agents are Everywhere: Will AI Shop and Pay for You?
With the proliferation of shopping agents, retailers are faced with the dilemma of choice. Often the decision revolves around organizational size. Rioux says many companies, particularly small ones, lack “the ecosystem or the brand awareness or just the patience of their shoppers” to merit a bespoke conversational agent.
Gilmore imagines a large retailer with multi-year investment in commerce for a new platform: “I own my reviews. I own my data, my product catalog, and that kind of thing. Do I invest in building my own agent or do I try to co-opt these third-party agents that are out there?”
Rioux replies that there is value in ownership for large retailers: “You get to learn a lot, and it also gives you the unfair advantage that you own the learning.”
Conversational agents, he says, enable them to “poke holes in your own inventory,” discover the questions consumers are asking and answer them with a speedy patch.
“Owning that estate and having that line of sight into your data landscape allows you to play a lot faster than if you either get something out of the box or [are] trying to play into a bigger ecosystem where your incentives are completely different.”
As for smaller retailers, Rioux says that bold ones might go with micro-agents.
“If you do fulfillment through a third party or you do commerce by a third party, then you’re doing yourself a disservice if you don’t play into that ecosystem.” But he adds that the rules are different here and that retailers must learn them and optimize differently.
4. Responsible Retail in the Age of AI: What are the Pitfalls?
Gilmore then wonders about the threat to retailers, of any size, over their brand when the agents take control.
Rioux says: “You're going to need to develop an entire new data product to be able to accommodate those agents.” And he says that you must “experiment, learn, be unapologetic about the results that you're getting and iterate fast.”
Note: It’s critical to remember the privacy and compliance issues involved when you’re speeding along in the AI lane. You need to be responsible when using AI. And it’s necessary to employ people who deeply understand what responsibility means. You need to mitigate risk against agents saying and doing the wrong thing. “Having a good partner, like EPAM, with a toolset and a framework to help you experiment responsibly and at the same time quickly,” says Gilmore. “Those kinds of things are key.”
Of course, not nearly enough organizations are properly set up here. According to our AI report, just 4% of the disruptors we surveyed have a full-fledged governance plan, compared to 1% of all the other participants. 1% of the rest of our participants.
5. The Consumer View: Trust, Transparency and Privacy Concerns
Now, our talking heads aren’t just talking heads; they are also consumers and so they have personal opinions about AI agents.
Even as a consumer, perhaps especially as one, Rioux believes in responsibility. He’s willing to share his data with brands that “are going to use my data responsibly, that are offering that transparency.”
And he’s not the only consumer who feels this way.
“Customers don't want to have the impression that they've been tricked,” says Gilmore. This puts real pressure on retailers and brands and companies to do the right thing.
6. The Future of Commerce: What about Next Year?
Rioux has much to say about where we are right now in the world of AI agents… but he wonders if Gilmore has thoughts about what the field could look like in 2026.
He has two responses.
One is from the perspective of retailers: “If you're a retailer, having a shopping agent that can create [a] relationship with the customer will be the new driver for revenue going forward.” Which means that retailers will need to have adopted the right platforms and composable architecture to enable such relationships.
As a consumer, he says: “I almost want to be able to create an agent for myself.” Gilmore’s 2026 agent would, using the “body of knowledge about Brian and what Brian likes to do,” of course make personalized shopping recommendations… “but at the same time, it could be recommending a trip for me or recommending non-transactional things for me to enrich my life.”
Watch the full video here.
