Merchants attending PLS in Amsterdam see AI-driven shopping as a long-term game changer

There were several lively discussions with merchants at Payment Leaders’ Summit (PLS) in Amsterdam which illuminated the sector’s latest challenges and opportunities with Artificial Intelligence (AI) and agentic commerce. The event brought together key stakeholders from retail, hospitality, and e-commerce, and opened with candid exchanges around the evolving payment preferences of consumers, growing demand for secure omnichannel solutions, and merchants’ strategies for navigating regulatory changes. These are the customary topics of many of the PLS events. However, this year, there was a consensus reached on the urgent need for closer collaboration between solution providers and merchants to drive seamless, customer-centric innovation in the rapidly changing AI landscape.

It was described during the many merchant roundtables that AI in retailing refers to the use of artificial intelligence technologies, such as machine learning and automation, to optimize retail operations, payment optimisation, enhancing personalised customer experiences, and improving decision-making based on data and predictive analytics in fraud prevention.

Speaking with merchants, it was discovered that some of them are already receiving requests from AI agents. Around 30% of merchants expect to receive traffic from AI agents within the next six months and 60% of merchants expect to receive traffic from AI agents within the next 12 months. Admittedly, this was not a robust or scientific survey, nevertheless it was based on conversations with over 20 merchant attendees from a wide selection of sectors – including, OTAs, airlines, hotels, and fashion and general merchandise retailers.

Security vs. AI fraud

When consumers use autonomous digital assistants, or AI agents, to search for products, compare offers, and make purchases on their behalf, it requires the merchant’s storefront, product data, APIs, and checkout flows to be optimised for AI agent-driven transactions. An important question was raised – what about strong customer authentication (SCA)? In agentic commerce, SCA is expected to be managed through a combination of tokenization, unique agent-customer interaction, and adaptive security protocols. Many of the merchants at the PLS didn’t know how AI agents use secure payment tokens or were simply unprepared for AI agents. Network tokens, generated and maintained by the international card schemes, link a specific agent to a customer, ensuring every transaction is both authenticated and authorized, with methods such as biometrics, passkeys, or one-time passwords adding further validation before purchase.

It was acknowledged that fraudsters exploiting AI agents to use real customers’ payment methods or Alternative Payment Methods outside the card scheme’s secure ecosystem pose a serious threat. What is the solution? No merchant at the PLS had a solution, only the expectation that AI agents will allow another route for the fraudsters to gain access and their fraud prevention measures need to be prepared.

Within Edgar, Dunn & Company’s Retail & Hospitality Practice it is expected that AI Agents may have to be registered and verified, using processes like “Know Your Customer” (KYC) for businesses – it will be effectively a KYA protocol - “Know Your Agent”. This will ensure only legitimate, traceable parties are permitted to transact. Any AI agent acting on behalf of a consumer must be authenticated and tied to a specific customer, making unauthorised agent behaviour easier to spot and block. The card scheme networks and Alternative Payment Methods are increasingly using tokenization, linking tokens not just to payment credentials but also AI agent identity, device characteristics, and real-time metadata. This makes stolen tokens or payment credentials much harder for fraudsters to use without matching agent and transaction context.

Transactions from AI agents will trigger continuous identity verification (e.g., biometrics, passkeys, dynamic risk scoring) and real-time monitoring of agent behaviour, device usage, and spending patterns. Unlike static checks, these dynamic protocols are expected to catch anomalies that reveal fraudulent use even for payments outside the card rails. However, can an AI complete a biometric check? AI-driven systems now perform biometric authentication, such as facial recognition, fingerprint, iris, and voice verification, by analysing biometric data, and confirming user identity for payment authorisation. Consequently, the fraudsters are expected to use the same AI tools to emulate biometric security seen in the payment industry. This was a serious concern amongst the merchants participating in several of the PLS roundtables.

Model Context Protocol (MCP) the communication standard for AI

MCP (Model Context Protocol) in agentic commerce serves as a standardised integration layer or protocol that enables AI agents to securely interact with external business systems, APIs, and data in a contextual, dynamic, and composable manner without requiring custom or bespoke API integrations. The merchants at the PLS along with the PSPs, such as Stripe, Adyen, and so on, are already making their platforms MCP-ready. MCP provides a common language and structure for AI agents to understand and trigger real business workflows, including browsing shop catalogues, managing shopping carts, checking inventory, processing payments, seamlessly. We heard from one hotel merchant at the PLS that AI agents are already making reservations for VIP customers who had previously registered with them. They were not MCP-ready, but it is expected that they will be within the next six months. Merchants broadly recognise that being MCP-ready will be crucial for thriving in the era of agentic commerce, and accordingly, are making it a strategic focus on their development agendas to ensure competitiveness as AI agents are poised for widespread adoption among shoppers.

Bad transactions from bad AI Agents

Agentic commerce is an exciting development in e-commerce for both the merchants and the payment service providers. Especially when the expectation is for more sales revenue and better checkout conversion. This is fine when the non-human transactions are good transactions. However, fraud from AI agents is a major concern amongst the merchants at the PLS.  It is expected that AI-driven fraud will account for a significant portion of total fraud, and it is expected to increase. Just as it was seen when plastic payment cards were protected with Chip and a PIN at the POS the fraudsters shifted their focus towards e-commerce. As AI agents become the norm to distinguish between good and bad AI agents will be a challenge.  

The merchants worry about sophisticated attacks using AI to mimic legitimate transactions, exploit AI shopping agents, and conduct large-scale fraud. In response, merchants and risk management companies are partnering to develop specialised tools that differentiate genuine AI-driven purchases from fraudulent ones, helping merchants safely embrace agentic commerce while mitigating the risks. Liability for fraud increasingly shifts to agent providers or platforms if they fail to secure AI agents. Rigorous standardisation around agent data-sharing and transaction logging is an area that will have huge implications for dispute resolution and investigation when fraud occurs. This was an area that none of the merchants knew whether they had an answer or simply unwilling to discuss at this payments industry conference.

The Future of Agentic Commerce

All the merchants at the PLS view agentic commerce as transformative, with the potential to enhance personalisation, improve the customer experience and payment optimisation. AI is expected to even speed up product development – this was mentioned by one Cloud Property Management System (PMS) attending the summit. However, merchants also recognise the need to strategically manage their data and the emerging risks. The expectations of agentic commerce are high, with strong returns and cost savings for merchants. Edgar, Dunn & Company (EDC) do not claim to have all the answers to the future of Agentic Commerce, nevertheless EDC will continue to work with merchants and PSPs to develop a roadmap that is designed to help merchants navigate the new world of agentic commerce.

The content of this article does not reflect the official opinion of Edgar, Dunn & Company. The information and views expressed in this publication belong solely to the author(s).