Foundational thinking on payments infrastructure, credit architecture, and the evolution of modern financial services.
33 articles
Issuers face a fork: be the passive endpoint of an agent's decision, or the layer that governs it.
The agent had permission and still got it wrong — a failure mode our rails have no category for.
Wall Street's blowout quarter was about money moving — not consumers borrowing more. Spending is up; borrowing is flat. That decoupling is where the next few years of lending and payments strategy get decided.
Agentic commerce isn't a faster checkout. It's a different actor at the checkout.
AI-driven memory scarcity is already pushing consumer hardware prices higher—forcing banks and issuers to choose between modernizing their origination and decisioning infrastructure or ceding the financing layer to faster competitors.
As AI agents execute transactions on behalf of consumers and businesses, the payment industry must move beyond narrow technical authorization to operationalize legal and regulatory proof of delegated authority. The question of what a user actually authorized an agent to do—and who is liable when outcomes diverge from intent—will determine whether agentic commerce scales.
Visa's announcements at the Payments Forum apply its sixty-year-old tollbooth playbook to agentic commerce: don't pick the AI winner, become the trust layer every transaction has to clear through. The standards for agent-driven commerce are being written this week — with you at the table or without you.
Technology decisions have become enterprise strategy decisions for community banks. Without integrated leadership between strategy and technology executives, institutions risk gradual erosion of competitive position through fragmented capabilities, slower product delivery, and reduced strategic optionality.
For many people, generosity becomes personal the moment a cause stops feeling abstract. Americans donated nearly $600 billion to charitable causes in 2024 — much of it through systems most people barely notice. Tiny transactions. Massive impact.
Walking through a college campus, watching students move fluidly through their day — coffee, meals, vending machines, no friction — one thought kept resurfacing. When payments work well, they disappear. Sometimes the deeper purpose of infrastructure becomes visible only in the moments where removing friction genuinely changes a human experience.
We often talk about platform transformation in payments as if it were something entirely new. It isn't. What's changed isn't the existence of transformation — it's the realization that front-end innovation depends on what the processing architecture underneath can actually support safely and flexibly at scale.
AI agents are no longer simply assisting consumers. They're increasingly executing payments autonomously on their behalf. Not hypothetically. In present tense. That changes the architecture of payments itself — and forces a deeper question about whether infrastructure can support that level of machine autonomy safely.
Lately, I've been thinking about program management in card issuing. And one question keeps surfacing — who actually controls the business? Whose logo is on the card? Who carries the credit risk? Execution can be delegated. Control cannot — at least, not without meaningful long-term trade-offs.
After visiting more than a dozen university campuses with my children over the years, I didn't expect to notice something fundamentally new. But this weekend at the University of Wisconsin–Madison, one observation became impossible to ignore. Payments simply worked — and that wasn't an accident. It was a preview.
Are legacy platforms genuinely becoming a structural constraint, or is that narrative overstated? Research from McKinsey, Accenture, Deloitte, BCG, and EY arrives at remarkably similar tensions. The question is no longer whether legacy systems work. It's whether they can stretch far enough to support a fundamentally different model of credit.
The premium card market has become one of the most competitive segments in payments. But rewards alone are not where the true differentiation resides. Increasingly, value is measured through the totality of the relationship — trust, service consistency, transparency, and how the institution behaves during the moments that matter most.
After 30+ years across payments, one observation stands out clearly. The economics behind the system are often deeply misunderstood. Many of the things commonly labeled as 'fees' are actually economic mechanisms that enable the ecosystem itself to function — and reducing them in isolation can create unintended consequences elsewhere.
A growing wave of state-level legislation is targeting interchange economics. Individually, the proposals pursue understandable objectives. Collectively, they introduce a different kind of challenge — because payments infrastructure doesn't fundamentally operate at the state level. Regulatory fragmentation may begin influencing not only the economics of payments, but the architecture itself.
Something deeper is shifting in how regional banks and credit unions approach infrastructure modernization. This is no longer simply about selecting a processor. Increasingly, it's about regaining control of the stack itself — and reclaiming the strategic authority that decades of outsourcing slowly distributed away.
Over the past 40 years, the number of banks and credit unions in the United States has declined dramatically. Yet the institutions that remain are larger, more technologically capable, and operating inside an increasingly complex digital ecosystem. Perhaps community finance isn't disappearing at all. Perhaps it's being reinvented.
The invention that changed the way we conduct commerce — and with it the way we live our daily lives — was not ultimately a piece of technology. Paradoxically, the reason electronic payments transformed society so profoundly was something much simpler. Trust embedded directly into the architecture of commerce.
Loyalty programs are usually framed as a marketing challenge — points, offers, campaigns, promotions. But the more I observe how the most successful programs actually operate, the more I believe the real differentiator exists somewhere else entirely. Loyalty is quietly becoming an issuing capability.
Most marketing mailers go directly into the discard pile. This one didn't. Aven's HELOC proposition isn't just a marketing innovation — it's fundamentally a processing innovation. A traditionally episodic lending product is becoming transactional, and that's a much deeper shift than it first appears.
BNPL is often discussed as a checkout experience or a UX innovation. But increasingly, the market is beginning to price it for what it fundamentally is. Credit. Yesterday's market activity, paired with New York's proposed regulatory framework, reinforced that distinction clearly.
Frictionless is not the end state. The larger shift underway is toward experiences that dissolve entirely — where credit no longer feels like a standalone product, but behaves like a capability embedded directly into the customer's moment of need. Architecture flexibility is becoming a prerequisite for relevance.
Issuers do not fundamentally have a UX problem. Increasingly, they have an architecture problem. Younger consumers expect credit to appear contextually, instantly, within the transaction flow itself. From the outside, it looks like UX evolution. Underneath, the shift is much deeper — and it's not marketing. It's platform architecture.
Jumping from a BNPL conversation into SMB lending might seem like a stretch. In reality, it isn't. Small and mid-sized businesses rarely struggle with credit conceptually — they struggle with timing. That's fundamentally a cash-flow coordination problem, and payments-anchored installment structures address it directly.
BNPL often gets discussed primarily as a consumer experience decision. In reality, it's frequently a platform and execution decision long before it becomes a user-experience discussion. Three different operating models — merchant-funded, issuer-funded, third-party — each place very different demands on the infrastructure stack.
Whether a transaction can be posted as deferred, split into installments, or follow distinct billing logic isn't determined at checkout. It's determined by what the processing platform can express after authorization. BNPL capabilities themselves are increasingly table stakes — the real differentiator is execution.
BNPL didn't arrive late in the U.S. — the conditions arrived late. I've been familiar with installment-based credit since the early 1990s, working across Brazil and Mexico. The traditional U.S. revolving-credit model functioned effectively for decades, until the surrounding economic conditions began changing materially.
BNPL sits directly in the gap between how consumers experience credit and how institutions traditionally model it. The consumer-side behavioral pattern is consistent and well-documented. But the issuer-side value proposition is discussed far less often — even though it may be just as important.
Most credit models assume people think in terms of credit limits, utilization, APRs, and revolving balances. But consumer behavior tells a very different story. Households experience credit through a much more immediate and practical lens — and that disconnect shapes everything from product design to underwriting logic.
I've heard predictions about a fully cashless future more times than I can count — and year after year, the reality on the ground tells a more nuanced story. Cash still represents a material share of point-of-sale transactions globally. What's changed isn't the relevance of ATMs. It's how they're built and managed.
Return to the main Insights page for the latest perspectives on payments and financial infrastructure.
View recent insights