Agentic commerce: why OpenAI's checkout announcement is not a setback

Chris Jones, Managing Director at PSE Consulting, explains why ChatGPT’s pause highlights structural constraints in agentic commerce and not a retreat.

The news that OpenAI has scaled back its Instant Checkout feature has attracted considerable commentary, much of it framing the move as a setback for agentic commerce. That reading misses the point. What we are witnessing is not a failure of the concept but a candid signal of just how complex it is to bring autonomous AI agents into real-world shopping environments at scale. These systems must navigate an extraordinarily broad landscape: product data, identity verification, fraud detection, tax compliance, liability frameworks, loyalty programmes, and payment flows, all coordinated seamlessly across millions of merchants. The ambition is enormous, and the infrastructure needed to support it is still being built.

The five constraints the industry must confront

A recent report from Adyen on agentic commerce makes this structural complexity impossible to ignore. Drawing on insights from more than 200 enterprise merchants and direct work with AI commerce platforms, it identifies five constraints that no single company can resolve alone: protocol fragmentation, product data not designed for machine consumption, legacy enterprise stacks, trust and liability frameworks, and onboarding at scale. Each of these is a significant challenge in its own right. Together, they create an environment in which solutions that perform impressively in demonstrations can break down quickly when deployed in production. The gap between what AI agents can theoretically do and what they can reliably do today is largely a function of these systemic limitations, not of the AI itself.

Protocol fragmentation alone presents a formidable barrier. Agentic commerce requires consistent, machine-readable communication between AI platforms and merchant systems. Today, those standards are absent or inconsistent. Product data compounds the problem: catalogues built for human browsing, with incomplete attributes, inconsistent formatting, and unreliable availability information, are poorly suited to autonomous decision-making. Add in the structural complexity of enterprise commerce stacks, many of which are built on legacy technology accumulated over decades, and the scale of the challenge becomes clear.

Where payments can move faster

Not all of these constraints are equally difficult to address, and it is worth being precise about where progress is achievable in the near term. From a payments perspective, several of the most pressing challenges are tractable through collaboration across payment networks, infrastructure providers, and AI platforms. Fraud detection, tax calculation, and payment flow optimisation are areas where scale, shared standards, and strategic partnerships can drive meaningful progress. These are not easy problems, but they are problems that the payments industry has experience solving, and the tools and incentives to do so already exist.

The same cannot be said for every dimension of agentic commerce. SKU-level accuracy, shopper identity management, real-time product availability, and the deep structural complexity of enterprise commerce stacks are embedded in how ecommerce functions today. Resolving them requires sustained coordination across merchants, platforms, and infrastructure providers over an extended period. Progress will come, but it will be measured in years rather than quarters.

Separating cause from effect

One of the most important distinctions to draw in discussions of agentic commerce is between cause and effect. There is a tendency to interpret merchant caution around enabling agentic purchasing as evidence that consumers are not interested. The evidence does not support that interpretation. Consumer appetite for the convenience of agent-driven commerce is genuine. The hesitation among merchants reflects something different: a recognition that the underlying systems are not yet ready to support that experience reliably and safely.

This distinction matters because it shapes how the industry should respond. The priority is not to manufacture demand or to persuade sceptical merchants. It is to build the foundations that will allow agentic commerce to function as intended. As those foundations are strengthened, the convenience of agentic-enabled shopping is likely to prove a powerful driver of wider adoption. The friction currently embedded in agentic commerce flows is a problem of infrastructure, not of appetite.

For payments specialists, the lesson from recent developments at OpenAI and from the Adyen research is consistent and clear. Agentic commerce cannot be solved through a payments lens alone. Success requires a holistic approach: systems that handle multiple protocols, maintain persistent and verified identity, carry accurate and machine-readable product data, manage liability transparently, and scale onboarding efficiently. These are not problems that any single player in the ecosystem can resolve in isolation. The current moment should not be read as a retreat. It is a reminder that the industry is still constructing the infrastructure required to make autonomous shopping safe, reliable and genuinely seamless, and that the work of building it is the defining challenge of this phase of commerce.

About the author

Chris Jones manages PSE Consulting’s business and is well known in the UK and EU for his regular insights into payments innovation. He has spent the last 20+ years leading assignments for major clients during his time at PSE and Accenture. His specialisations include customer proposition development, market entry strategies, and enterprise value creation. Chris is a highly effective communicator, with very strong analytical skills and able to deliver recommendations to C-level clients and company boards. Chris regularly supports major enterprises, corporates, and global digital merchants’ payments, bringing new market perspectives and identifying fresh opportunities for innovation and expansion. He has also delivered payment assignments for fintechs, banks, and processors on topics such as regulatory impacts, new acceptance methods, open banking, BNPL, gateway/acquirer convergence, and opportunities for M&A and inorganic growth.

About PSE Consulting

PSE Consulting is a leading global provider of payment advisory services to players across the payments landscape. PSE’s expertise has enabled it to deliver actionable market insights and operational optimisation to senior payments leaders for over 30 years. To learn more, visit: https://pseconsulting.com/