Ecommpay webinar write-up: who builds trust in the age of agentic commerce?

Paula Albu, Junior Editor at The Paypers, shares the key insights from Ecommpay’s webinar Who Builds Trust in the Age of Agentic Commerce?, covering consumer trust, AI agents, liability frameworks, and the future of agentic commerce.

Agentic commerce is generating intense interest across the payment industry, moving from an industry buzzword to experimentation. Ecommpay’s latest research, based on interviews with more than 1,500 European consumers and 16 merchants and payment intermediaries, reveals a significant trust gap that must be closed before agentic transactions can scale.  

In a recent webinar hosted by The Paypers and Ecommpay, industry experts discussed the opportunities, risks, and infrastructure requirements that would shape the next phase of digital commerce. Esther Groen, NED, Strategy Consultancy, Banking & Fintech, Financial Health & Inclusion, and Chair moderated the discussion.  

The panel featured specialists Roy Blokker, Head of Strategic Sales at Ecommpay, Chris Jones, Managing Director, PSE Consultants, Shyam Sreenivasan, Solution Architect, ePayments at Würth Group, and Orestis Papasternos, Senior Product Manager at Booking.com. 

The trust paradox: consumers want AI help, but not AI control 

Ecommpay’s research shows that consumer appetite for agentic commerce is already significant, with 74% of responders expressing interest in AI-powered shopping experiences. They see the value in practical tasks such as comparing prices, finding discounts, recommending products based on preferences, adding items to baskets, and completing purchases.  

However, trust remains conditional. 21.3% of consumers would not allow an AI agent to spend money on their behalf, while 44.3% would limit spending to EUR 100. Only 6% would permit agents to spend more than EUR 500. 

For the panel, these findings reinforced a critical insight: consumers are interested in delegation, but not in surrendering control. Chris emphasized that the market is still in its baby steps, with most experimentation happening in a human-in-the-loop model rather than with full autonomy. 

 ‘’Know Your Agent’’ could become as important as KYC 

A recurring theme throughout the discussion was the need for greater visibility into agent identity and behaviour. 

Several speakers argued that future commerce ecosystems may require a ‘’Know Your Agent’’ (KYA) approach similar to existing Know Your Customer (KYC) requirements. Shyam highlighted that in B2B environments, this becomes even more critical, as agents may operate across procurement, recurring purchasing, and multi-entity corporate structures.  

Merchants need confidence not only in who the customer is, but also in which agent is acting on their behalf, what permissions have been granted, and whether those permissions are being respected.  

For merchants, this means understanding: 

• Which agent is initiating the transaction; 

• What spending limits apply; 

• What actions have been authorised; 

• And how decisions can be audited and explained. 

Moreover, one of the key concerns underpinning the entire discussion was the evolving fraud and liability landscape in an agent-driven environment. Traditional e-commerce fraud detection models rely heavily on human behavioural signals such as device fingerprints, typing patterns, or session tracking, all of which become significantly less reliable when transactions are initiated by autonomous agents. As Orestis pointed out, this creates a gap between what an agent is technically authorised to do and what a customer can later prove they approved. As a result, liability attribution becomes increasingly complex, reinforcing the need for clear frameworks. 

Visible control, not full autonomy  

Despite the revenue around autonomous purchasing, the panel consistently returned to one idea: humans must remain part of the decision-making process.  

Orestis highlighted several mechanisms already being explored, including spending limits, approval prompts, and human-in-the-loop controls that allow users to review and approve actions before transactions are completed. 

Rather than removing consumers from the purchasing journey, successful agentic commerce models are likely to focus on making delegation transparent and controllable.  

As Roy observed, trust is built through visible control mechanisms, not through eliminating customer involvement.  

PSPs could become the trust layer of agentic commerce.  

One of the most interesting conclusions from the webinar was the potential role of payment service providers.  

As Roy mentioned, PSPs already sit at the intersection of merchants, consumers, issuers, and payment networks; they are well-positioned to provide the visibility, authentication, and monitoring capabilities required for agentic commerce. 

PSPs could help merchants understand which agents are interacting with their platforms, provide information about permissions and authorisations, and support experimentation while protecting brand trust.  

Experimentation stage: early but accelerating 

Despite growing interest, agentic commerce traffic remains relatively small today. However, panelists agree that adoption is increasing and that merchants should begin preparing now.  

The immediate priority is not large-scale deployment but learning. Businesses should start experimenting, ensure that product and pricing data are machine-readable, establish clear controls, and work closely with payment providers to understand emerging protocols and standards. 

The speakers’ final message was consistent: the future of agentic commerce will not be determined by technology alone. Success will demand building trust through transparency, accountability, and clear human oversight.