Nuvei Says Merchants Need Their Own AI Shopping Agents

Highlights

Agent errors leave merchants and consumers seeking clarity on liability.

Retailer-owned agents give sellers more say over the shopping experience.

Nuvei’s payment test kept shopper approval before automated checkout.

Watch more: The Prompt Economy With Advait Sinha of Nuvei

    Get the Full Story

    Complete the form to unlock this article and enjoy unlimited free access to all PYMNTS content — no additional logins required.

    Subscribe to our daily newsletter, PYMNTS Today.

    By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions.

    Merchants preparing to sell to artificial intelligence (AI) agents have two decisions to make: how much of the customer relationship they’ll surrender, and how much responsibility they’ll accept when an agent buys something its owner didn’t want.

    The commercial appeal of agentic commerce must pair with the trust required to sustain it. Advait Sinha, global head of product at Nuvei, said that the biggest hurdle to agentic adoption is establishing trust across the payments ecosystem.

    Existing payment practices provide procedures for tracing transactions, identifying fraud and assessing chargeback evidence. Delegated purchasing introduces another participant whose instructions and actions need to be understood.

    “What is the protection that’s offered both to the consumers as well as the merchants in those situations?” Sinha told PYMNTS.

    He offered up examples of purchases that fail to meet ordinary expectations: the wrong shoe size or an item that won’t ship for two months. Improving language models should reduce errors, he said, but mistakes will still occur. Consumers and merchants need confidence in the arrangements for handling them.

    “Technology can move much faster than habits,” he said. In order to bridge the gap between the tech that’s available and the actual use of that technology in everyday financial life, trust must be reinforced with each interaction.  Trust builds confidence, which in turn will lead to a wider adoption of agentic commerce.

    Some of the foundations of trust have already been firmly established. Consumers already use AI to evaluate reviews, compare prices and examine quality, warranties and support. Authorizing software to spend money requires additional confidence, as well as payment infrastructure prepared to process those purchases.

    Nuvei tested that sequence with Visa, Arvato Systems and merchant Kings and Priests. An agent embedded in the merchant’s website responded to a shopper’s request and proposed an item.

    “The shopper approved it, and then everything after that was the agent acting alone,” Sinha noted.

    He described the subsequent steps as completing the purchase, obtaining payment approval and notifying the customer that the item was due to ship in two days. The demonstration retained human approval of the selection while delegating the work that followed.

    It also placed the agent inside the merchant’s own shopping environment, a choice with consequences for how sellers participate in agentic commerce.

    Who Owns the Shopping Experience

    Sinha said the prominence of general-purpose AI providers has clouded the discussion of how agent-led shopping might develop.

    “We get lost because everyone talks about your Claude, your OpenAI, and the established open-market agents when it comes to some of these things,” he said.

    He sees room for agents built around particular industries and individual merchants. That leaves sellers with choices about where to introduce their own capabilities and how to work with agents arriving from elsewhere.

    “This technology story has not been written yet,” he said.

    A merchant whose products are selected through an outside service has fewer opportunities to influence their presentation or explain related offers. Its relationship with the buyer changes when another company supplies the shopping interface.

    “Your brand is not as prominent and you don’t control the customer experience,” Sinha said.

    Nuvei advises merchants to consider developing their own agents before deciding how to participate with general-purpose alternatives. Sinha cited Expedia‘s approach to illustrate the appeal: Travel involves several connected decisions, creating opportunities to customize recommendations and introduce additional services during the purchase process.

    For payment providers, preparing for those transactions is one part of a broader AI agenda. Sinha described three applications at Nuvei: internal productivity, AI within merchant-facing products and services, and adapting those services for agentic commerce.

    The internal work includes employee tools and processes performed by agents. Merchant applications include payment decision-making and interactive dashboards for examining transactions and revenue. Sinha identified better approvals and merchant interaction as objectives, without supplying measured results.

    Nuvei is also working with Visa and Mastercard on the arrangements supporting agent-initiated payments. Sinha described wider participation as essential before delegated checkout becomes routine.

    “Everyone needs some time to catch up before the full ecosystem is ready,” he said.

    Watch the full interview with Advait Sinha to learn more about:

    • How a shopper’s approval led to an agent completing Nuvei’s Visa payment demonstration.
    • Why merchant-operated agents could support more tailored travel purchases.
    • Where Nuvei sees applications for AI within its operations and merchant products.

     

    Advait Sinha is global head of product at Nuvei.