"Design for a world where your user isn’t always human": A Conversation with Ben Smith from Stripe

From enterprise systems at Broadcom to cloud-native architectures at AWS and now payments at Stripe, Ben Smith has built a career around one core idea: making complex systems understandable and usable for developers.

Through hands-on demos, workshops, and real-world simulations, he focuses on bridging the gap between how systems are designed and how they actually behave in production.

In the interview below, he talks about fintech’s biggest blind spots, the shift toward agentic commerce, and why the future belongs to developers who build complete systems, not just integrations.

🧭 Turning Complex Systems into Real Developer Experiences

For Ben, every career move has followed the same direction: turning abstraction into something tangible. “For me, it’s about making complex systems real and usable for developers. At Broadcom, I was working on automation and integrations across enterprise systems, getting a grounding in how real-world processes actually run. At AWS, that evolved into cloud-native systems, serverless and event-driven architectures, and I focused on making those tangible through things like Serverlesspresso, which is a full end-to-end coffee ordering system that simulates a real business.”

What matters is not just what systems do, but how they behave. “Developers can see how orders flow through events, how systems coordinate, and how everything behaves under load. At Stripe, I’m doing the same thing for payments, building hands-on experiences like the Swag Store demo, and now launching agentic commerce workshops, where developers can explore how AI agents actually initiate and manage transactions.”

 

🔁 Bridging Technical Systems and Business Reality

That focus on real systems naturally shaped how he connects engineering work to outcomes.

“When you build real systems, not just isolated features, you’re forced to think about how users interact with them, what outcomes they drive, and where things break. At AWS, I was feeding structured customer insights back into product teams and seeing how that shaped the roadmap. At Stripe, I’m working across product and marketing to turn things like Workflows or agentic commerce into something developers actually care about.”

Over time, that translation becomes less about explaining, and more about meaning. “Over time, you learn that translation is really about connecting what you’re building to what it enables.”

⚙️ Where Fintech Still Falls Short for Developers

Despite strong APIs and mature platforms, something critical is still missing.

“Fintech still gives developers APIs, but not enough understanding of how things behave in reality. The gaps I see are unclear async and failure states, fragmented global behavior, limited ways to experiment safely with real flows.”

The real shift happens when developers stop seeing pieces—and start seeing systems.

“From my experience, the biggest unlock is when developers can see the whole system. That’s why I focus on building demos and workshops that are end-to-end, because once you can trigger a payment, see the events, and understand what happens next, everything clicks much faster.”

🤖 Rethinking Architecture for Agentic Commerce

As AI agents begin to interact directly with systems, many teams are still designing with outdated assumptions. “They’re still designing around human interaction patterns. Most systems assume a user initiates something, the system responds, and the flow ends. But agents don’t behave like that, they act over time, across systems, and often asynchronously.”

This creates a fundamental mismatch between design and reality.

“The mistake is not designing for systems that operate continuously, decisions that evolve, interactions that aren’t tightly coupled to a UI. This is where patterns from event-driven and distributed systems become really important, but applied in a new context.” 

⚖️ Responsibility in AI-Driven Transactions

As systems gain autonomy, responsibility becomes harder to define.

“It’s going to be shared responsibility, but we don’t have the right abstractions for it yet. Right now, the user sets intent, the agent interprets it, and the platform executes.”

The real challenge lies in what’s missing between those layers. “What’s missing is clear delegation boundaries, auditability of decisions, better ways to unwind or correct transactions. This is something I’m actively exploring with the agentic commerce work, because once agents are transacting, these edge cases become the main problem.”

🔍 Building Trust Without Visibility

In a world where users no longer see every interaction, trust needs to be designed differently. “You replace visibility with confidence. That means clear limits on what the system can do, predictable behaviour, the ability to inspect what happened after the fact.

This shift isn’t entirely new, it’s an evolution.

“We’ve already seen this shift with cloud and serverless, developers trust systems they don’t see because they behave consistently. The same principle applies here.”

🎨 Designing for a Non-Human User

Looking ahead, the biggest mindset shift is already happening. “Design for a world where your user isn’t always human. That means building APIs and systems, not just UI-driven flows, making everything resilient, retryable and idempotent, thinking about intent and execution as separate concerns.”

And in the end, what separates good developers from great ones is execution.

“And most importantly: Build real things. The developers who stand out are the ones who don’t just integrate APIs, they show how everything works together in practice. That’s what makes people understand, and that’s what drives adoption.”

💡 Final Takeaway

In an increasingly abstracted world, the real advantage doesn’t come from access to tools, but from making systems understandable end-to-end. 

🎫 Meet Ben Smith at Bucharest Tech Week

Whether you're building fintech products, exploring AI-driven commerce, or designing distributed systems, Ben Smith brings a practical perspective grounded in real-world implementation.

📍 Join him live at the AI Coding Summit on 18 June at Nord Events Center by GlobalWorth.