Jason Sharples, SVP and Unit CIO of Global Merchant and Network Services, American Express – Interview Series

Jason Sharples, SVP and Unit CIO of Global Merchant and Network Services, American Express, leads technology strategy and engineering in support of the company’s Global Merchant Services and Global Network Services businesses from New York. In this role, he is responsible for aligning large-scale engineering initiatives with business objectives, ensuring secure, resilient, and high-performing platforms that power merchant acceptance and global network capabilities.

Prior to American Express, he spent more than a decade in senior leadership roles at Global Payments, including CTO of Engineering and Operations and Senior Vice President positions overseeing infrastructure, application delivery, and enterprise production environments. He directed global teams, shaped long-term architecture strategy, improved release cycles, and drove major technology investments supporting multi-million-dollar transformation programs. His earlier career spans regional IT leadership, large-scale system conversions across Asia-Pacific, and technical consulting roles, reflecting a deep foundation in enterprise infrastructure, operational reliability, and payments technology.

You’ve spent more than three decades moving from hands-on technical consulting roles into leading some of the most complex payment systems in the world. Looking back at your journey from early infrastructure and systems work through Global Payments and now at American Express, what experiences most shaped how you think about reliability, scale, and innovation in financial networks today?

Over more than three decades, I’ve had the benefit of seeing payments from almost every angle — from frontline production support and infrastructure testing to running global platforms across multiple continents. Early career lessons stick with you. When you’re on call at 2 a.m. because transactions aren’t authorizing, you learn very quickly that reliability isn’t theoretical.

Those early experiences shaped how I think about scale today. Innovation in payments only matters if it works at peak moments, such as Black Friday, and continues to work the next morning. As I’ve built my career first at Global Payments and now at American Express, I’ve seen firsthand that disciplined engineering, strong operational hygiene, and modular design enable innovation without destabilizing the network. In payments, success often looks uneventful.

American Express processes tens of millions of transactions daily. When you think about modernizing a payments network at that scale, what architectural principles matter most to ensure innovation doesn’t come at the cost of stability or trust?

When you’re modernizing a network that processes millions of transactions a day, the most important architectural principle is decoupling change from risk. That means modularity, reuse, and clear contracts between systems. Risk of change is further managed by layered observability across all aspects of processing and contracts.

At Amex, our approach to innovation is through continuous modernization of our foundational tech infrastructure. Through this approach and by building a component-based network that allows new capabilities to be introduced incrementally rather than through large, one-off transformations, our team is empowered to innovate quickly and at scale while preserving the seamless experience and predictability that merchants and cardmembers depend on. For instance, when we started exploring agentic capabilities, we were well-positioned because of our reusable components.

There’s a lot of hype around generative and agentic AI in financial services. Where are you seeing these technologies deliver measurable productivity or customer experience gains inside large payment organizations today?

The most immediate gains we’re seeing from generative AI are practical in nature. AI-assisted development, policy interpretation, documentation and knowledge discovery, and client-facing workflows are delivering productivity improvements across business and technology teams.

Amex is thoughtfully exploring agentic AI and its ability to shape how secure, autonomous transactions work across the payments ecosystem. For instance, we are working with industry leaders such as OpenAI, Stripe, and Microsoft as they develop pilots like ChatGPT Commerce and Copilot’s Consumer Shopping Pilot. While it’s still too early to assess the impact of these pilots, we’re excited about the potential they hold to enhance technological capabilities and improve customer experiences. We are aware of course that there remain known and unknown challenges to solve, similar to the rise of eCommerce payments infrastructure 20 years ago, which led to unanticipated use cases and capabilities that are taken for granted today.

Payments infrastructure has traditionally been deterministic and rules-based. How does introducing more autonomous or agent-driven systems change the way you approach authorization, risk management, and exception handling?

Payments infrastructure has always been rules-based for a reason: predictability, fairness, and auditability. Introducing agent-driven systems only augments that foundation.

We approach autonomy as a layered capability. Deterministic controls still anchor authorization, settlement, and risk decisions.  Eventually, agents might assist with optimization, exception handling, and decision support. The goal isn’t to remove control, but to apply intelligence where it reduces friction without increasing uncertainty. Importantly those layers might include the use of agents to monitor the autonomous actions of other agents.

As AI systems take on more decision-making responsibilities, how do you balance speed and automation with the need for explainability, auditability, and regulatory confidence?

As AI systems take on more responsibility, explainability becomes more important. Speed without accountability erodes trust in financial networks very quickly.

That’s why we aim to design AI systems with auditability and transparency built in from the start, using clear decision trails, human-override mechanisms where appropriate, and alignment with applicable regulatory expectations across jurisdictions. Automation should accelerate outcomes rather than obscure how they were achieved.

From your perspective, what are the biggest technical bottlenecks holding back truly real-time, intelligent commerce experiences for merchants and consumers?

Real-time experiences are possible, but are subject to balancing factors such as risk, trust, accuracy, and outcomes. Each of these factors benefits from incremental intelligence being applied to alleviate bottlenecks over time.

How does enterprise-grade reliability differ when you’re operating a global payments network versus other large-scale enterprise systems, and what lessons might other industries learn from payments?

When dealing with something as delicate as the movement of money, ensuring trust in every interaction is critical. This trust is obtained and retained through very high availability, accuracy and security, all monitored through layered observability, and backed by effective restoration and adjustment processes.  For industries with the same desired outcomes, I would recommend focusing on observability as the foundation.

With agentic commerce on the horizon, how do you envision the relationship between merchants, consumers, and payment networks evolving over the next few years? 

As commerce shifts from web pages to agents, I believe the relationship between merchants, consumers, and networks will evolve from transaction execution to outcome facilitation. It’s likely that networks will increasingly serve as trusted intermediaries that help agents negotiate, authenticate, and transact safely on behalf of humans.

For merchants, especially small businesses, this could lower the barriers to global reach. For consumers, it could reduce friction and cognitive load. And for networks, I think it could raise the bar on trust, identity, and governance: areas where experience and scale matter deeply.

Security and resilience are non-negotiable in payments. How does the rise of AI-driven infrastructure change the threat model, and what new risks are leaders underestimating?

AI changes the threat model in two directions at once. Defensively, it gives us better tools for anomaly detection, fraud prevention, and operational resilience. Offensively, it lowers the barrier for sophisticated attacks and increases the speed at which threats evolve.

One of the most underestimated risks is over-automation without sufficient guardrails, especially as systems that move faster than humans can increasingly understand and intervene. In payments, resilience comes from layered defenses, disciplined design, and a deep respect for the trust customers place in the network.

For investors and technologists watching the payments space closely, what signals should they look for to distinguish incremental AI adoption from truly transformative shifts in financial infrastructure?

All companies operating in financial infrastructure have a backlog of projects and product development to accomplish their strategic objectives. Signals of a transformative shift may be initially obscured, but will likely be made visible through consistently successful and increasingly rapid execution that stems from AI augmented decision making, governance and adaptability.

This transformation may generate more modular, composable capabilities that lay the foundation for an  application or service that quickly captures the imagination of consumers.

Thank you for the great interview.