Most merchants still start their day the same way: log in, check a dashboard, compare money in against money out. Mark Alsentzer, Chief Technology Officer at Kurv, argues that this decades-old ritual is exactly what AI should be eliminating, not improving.
Mark Alsentzer, Chief Technology Officer at Kurv, has spent his career moving across different corners of the payments industry, an experience he says shapes how he approaches technology decisions today. In his new role as CTO at Kurv, Alsentzer oversees product and services enablement, organizational transformation initiatives, and technology strategy. In this interview with PaySpace Magazine, Alsentzer discusses the shift from reactive to proactive AI, how Kurv is building data architecture to support AI features without compromising security, and where he expects generative AI to deliver the most near-term value for small and medium-sized businesses.
- You’ve had first-hand experience with different aspects of the payment business throughout your career. How has that end-to-end exposure to the payments industry shaped the way you approach the CTO role at Kurv?
Working in different areas of the payments industry has shown me that technology shapes the entire experience, not just what merchants see at the top of the funnel. It might be tempting to focus on the interface or dashboard, since that’s what everyone uses. But I’ve found that some of the biggest improvements happen behind the scenes, in areas like transaction routing, approval decisions, and overnight reconciliation.
When those behind-the-scenes systems work well, they naturally improve the experience merchants do see. As CTO at Kurv, I focus on both sides, using technology to improve the parts of payments merchants interact with and the parts they never see, but that ultimately make their experience better.
- Does your academic background help form the AI strategy you’re building into Kurv’s platform today? Or do you rely more on practical insights from leading technology strategy across different startups?
My academic background gives me a useful reference point, mostly because it’s a reminder that AI isn’t new. It’s been around in some form for over 60 years, and some foundational elements have remained the same. Those fundamentals help define the “edges of the sandbox,” or the things that hold true for any company building with AI, regardless of industry.
But that foundation only gets you so far. Every company has a different application for AI and a different definition of value. Payments are no exception. Some of my biggest learnings about how AI can simplify payments for merchants have come from hands-on experience leading technology strategy across different companies, not just from textbooks. That’s why I rely on both. The academic foundation helps me understand what stays consistent, while practical experience helps me figure out what makes sense for Kurv and its merchants.
- In your earlier statements, you spoke of the role of “proactive AI” in eliminating the need for other payment innovations. Can you elaborate on this idea in more detail?
AI transforms, and in many cases, has the potential to eliminate the legacy paradigms that have historically shaped payment behaviors, while removing manual steps from the payments process. Take reporting dashboards for instance. Every merchant has some version of this. They log in on a certain day, review the numbers, check money in against money out, and use that information to make decisions about their business.
It’s a standard practice that has been around for decades. Reactive AI takes that familiar behavior and makes the dashboard better, with cleaner visualizations and maybe some predictive insights layered on top. That’s useful, but the merchant still has to show up and review the information.
Proactive AI takes a different approach. Instead of building a better dashboard, it asks whether the dashboard is needed at all. Rather than making merchants log in and interpret the data themselves, the system can work with that information and take action within rules the merchant has set. It removes the manual step entirely.
It’s not just about making existing tools smarter. It’s about creating a new way of working that gives merchants time back to focus on running their businesses.
- Modern data infrastructure is often described as the foundation AI innovation is built on. How is Kurv approaching data architecture to support AI-driven features without compromising security or compliance?
That’s accurate, and I would frame it as the latest link in a chain that started with big data, moved through machine learning, and has now arrived at AI. None of these steps happened in isolation. AI is only as effective as the data it can access, learn from, and ultimately understand, so our architecture has to make sure the right data is available, clean, and connected.
But security always comes first for me. Efficiency with AI and data can have both positive and negative impacts. It can amplify strengths, but it can also expose security weaknesses much more efficiently. That’s why, before we build any AI solutions at Kurv, we start with security and compliance from day one. This is crucial because an error at scale can be far more consequential than a small, manual mistake.
- Small and medium-sized businesses often lack the resources of larger enterprises to evaluate or implement new payment technology. How do you design AI-driven tools at Kurv so they add value for SMBs without adding complexity?
One of the things I find genuinely exciting about this moment is that AI levels the playing field for small businesses in a way that wasn’t possible even a few years ago. An SMB with access to a modest AI tool can evaluate industry trends and make informed decisions just as effectively as a business with a massive analytics team and a much bigger budget. That changes what we build for. The AI-driven tools we’ve deployed at Kurv and those we’re continuing to build are designed to serve as an experience utility layer. They’re not meant to be another system the merchant has to learn or manage. They sit underneath and accelerate the tools merchants are already using, so the value shows up without adding another thing to configure or maintain. That’s the standard I hold every AI feature to: does it make things faster and simpler for a merchant with no dedicated tech team, or does it just add another dashboard to check?
- Looking ahead, what’s one prediction you’d make about how AI will reshape the merchant payments experience over the next two to three years?
Fraud reduction is the obvious answer, and it’ll keep improving, but it’s not the part I find most interesting. Some developments, such as Agentic Commerce and Autonomous Transactions, are already emerging. Over the next two to three years, I think the focus will move toward personalization, not just in how things look but in decisions made for each merchant. For example, a coffee shop and a construction company have very different risks, cash flow, and customers. AI should recognize these differences instead of treating everyone the same. I’ll also be watching new regulations closely. This is genuinely transformative technology being applied to real money moving in real time, and the regulatory landscape is going to have to catch up. That part of the story is far from settled, and it’s going to be one of the more interesting threads to follow.
- Kurv’s mission is “Modern Payments, Made Simple.” From a technical standpoint, what does “simple” for end users actually require under the hood? What are the biggest engineering challenges in making a complex payments stack feel invisible to SMB merchants?
Simple on the surface is one of the hardest things to engineer. Under the hood, it means abstracting away underwriting, tokenization, routing, compliance checks, and reconciliation into something that just works, without the merchant ever seeing the moving parts. The real challenge isn’t any single piece of that stack, it’s getting systems that were never built to talk to each other, like a POS, a gateway, a CRM, and the banking rails behind them, to actually share data cleanly. When that data doesn’t flow, merchants end up doing the integration work themselves, manually reconciling numbers across platforms, and that’s the opposite of simple. Managing complicated situations smoothly in the background is critical. If a card gets declined, a payment needs to be retried, or a transaction is flagged, the system should fix these problems without the merchant having to step in. The system also needs to stay reliable, even during busy times like the holiday rush.
- Beyond fraud management, where else in the payments stack, e.g. underwriting, reconciliation, customer support, chargebacks, or something else entirely, do you see generative AI delivering the most near-term value for SMBs?
I believe chargeback and dispute management is where we’ll see the biggest impact soon. For SMBs, gathering evidence to fight a dispute takes more time than checking records or even talking to customers. This is the kind of task generative AI handles well: collecting transaction logs, delivery confirmations, and communication history into a clear response, so merchants don’t have to do it themselves late at night.
- The Merchant Risk Council’s latest Global Payments & Fraud Report found that 56% of merchants now use generative AI for fraud management, up from 42% a year before. What’s driving that jump, and do you think AI-powered fraud prevention is already in its prime or is it just at its nascent stages?
That kind of jump usually comes down to accessibility more than any single breakthrough. A few years ago, using AI for fraud meant building or buying a custom model, which was out of reach for most SMBs.
Now, AI-based risk scoring is built into tools merchants already use, like gateways and processors, so it’s becoming a standard feature instead of something extra. Growing fraud losses and tighter profit margins are also driving this trend. Our own research found that 31% of SMB owners have faced AI-driven scams this year, with 20% reporting AI-generated invoices or payment requests convincing enough to pass for a real vendor, so the pressure to adopt smarter detection isn’t easing up anytime soon. Still, the industry is just getting started with AI adoption.
Right now, merchants mostly benefit from better detection, as AI models can spot suspicious patterns in real time. The next step will be automated decision-making, where systems both flag transactions and take action based on rules set by the merchant. This kind of agentic capability is still being developed.
- What’s a common misconception merchants or even industry peers have about how ready core payment operations are for AI, especially its agentic component?
The biggest misconception is that AI, especially the agentic kind, is a plug-and-play replacement for the systems merchants already have. It isn’t yet, and honestly it shouldn’t be. The bottleneck usually isn’t the model, it’s whether a merchant’s data is clean and connected enough for AI to act on with any confidence. A system can’t make good autonomous decisions with messy or siloed data without extensive cost implications, no matter how advanced it is. This doesn’t work for SMBs. The other misconception is around trust and control.
Many people think that agentic AI runs without any oversight. In reality, the businesses that get the most out of AI are the ones that combine it with their existing rules and workflows, setting clear limits on what the AI can do on its own and when a person needs to step in. This way is both more practical and safer than handing everything over to AI.
