Green Sheet interviews Treasury Prime's Chris Dean on smart AI deployment

As banks and fintechs race to integrate artificial intelligence into their operations, many are discovering that adoption alone doesn't guarantee meaningful results. In this Q&A, The Green Sheet checks in with Chris Dean, co-founder and CEO of Treasury Prime, about where AI initiatives most often fall short and what it takes to move from surface-level experimentation to real, scalable impact.

Dean shares insights on the architectural foundations required for effective AI, what genuine ROI looks like in embedded banking, how institutions can balance innovation with regulatory scrutiny, and why Treasury Prime believes generative AI is now mature enough to tackle one of embedded finance's most persistent bottlenecks.

Green Sheet: Many banks say they're using AI, but few seem satisfied with the outcomes. Where do you see the biggest disconnect between deployment and real effectiveness?

Chris Dean: AI is a change in how business is done, and things that weren't possible five years ago are possible now. Most banks start with low-risk tools like chatbots, but those bots rarely connect to real account data and rarely are as helpful as an expert human. They can sometimes answer basic questions, but they can't fix an issue for a customer. 

This highlights the core issue between deployment and real effectiveness: AI is only as useful as the system it's allowed to interact with. If the underlying architecture can't safely provide the right data with the right controls, AI will always sit on the surface instead of doing anything meaningful.

GS: AI pilots are one thing; scaling them across a regulated institution is another. What challenges do banks most often underestimate when moving from experimentation to production?

CD: Pilots are easy because they run in controlled settings. Production challenges banks to uphold accuracy, privacy and audits in real situations. The challenge banks most often underestimate is that this is a change in how the bank interacts with customers, data and money movement. An AI pilot can't grow without the right architecture underneath. 

GS: As pressure grows to demonstrate ROI, what does "real value" from AI actually look like in an embedded banking environment?

CD: Real value shows up when AI actually gets work done. That means completing a transaction, solving a customer problem or handling a routine task that used to take a person's time. In an embedded banking environment, this can look like auto-reconciling a transaction discrepancy or pre-building an onboarding checklist for a new fintech partner based on the bank's risk criteria.

Banks don't need AI to be clever; they need it to be useful. When AI can take bounded action on top of quality data and strong controls, the ROI becomes clear.

GS: With regulators taking a closer look at AI, how can banks and fintechs use it safely without stalling innovation or moving too fast without proper controls?

CD: Speed and safety aren't opposites; that's a false tradeoff. Banks and fintechs can move quickly if they build in strong oversight from the beginning. Since banks already know how to manage risk, the goal is to fit AI inside the frameworks they currently use. AI needs clear rules and humans overseeing the right things. With that foundation, you can innovate responsibly without creating unnecessary risk. 

GS: Looking ahead to 2026, what common misconceptions about AI in banking concern you most, even among institutions that believe they're ahead?

CD: A big misconception is treating AI like a "set it and forget it" system. It needs supervision, clean data and ongoing course-correction. Even the most advanced institutions underestimate the operational work required to keep AI aligned with compliance, customer expectations and the actual behavior of the underlying system.

GS: Treasury Prime is preparing to launch a generative AI–driven product aimed at a core embedded finance challenge. At a high level, what problem is it designed to solve, and why is generative AI the right fit now?

CD: Bank-fintech partnerships involve a lot of manual research, back-and-forth, and long diligence cycles. Like so many manual processes, it takes time and a lot of effort.

Our AI Marketplace uses generative AI to intelligently match banks with fintechs that fit their strategy and risk profile, and to streamline early evaluation and onboarding. It cuts through the noise by surfacing the right opportunities and giving both sides a clearer picture so they can make decisions faster and with greater confidence. By changing from a manual process to a mostly automated process we can get better answers faster.

Generative AI is the right technology to accelerate embedded finance collaboration because it now has the reasoning and language capabilities to understand complex partnership criteria, all while working inside the controls banks require. It's hard to imagine another technology providing the same efficiencies. 

Notice to readers: These are archived articles. Contact information, links and other details may be out of date. We regret any inconvenience.