With a distinguished career in banking and finance, Priya Jaiswal has become a leading voice on the intersection of technology and financial services. Her insights into market analysis and international business trends provide a crucial lens through which to view the industry’s rapid transformation. Today, we explore the pressing challenges and transformative opportunities presented by artificial intelligence, delving into the nuances of data strategy, the rise of agentic AI in payments, and the fundamental shifts in user experience that will define the next generation of financial institutions.
You noted the industry’s long-standing challenge of creating a “golden record” of data. Considering the pace of AI, could you walk us through the step-by-step process of how an institution can use an AI platform as a “forcing factor” to solve data integrity issues, rather than delaying?
For years, we’ve been chasing this phantom, this perfect “golden record,” trying to massage every piece of data across the organization into a flawless 360-degree view. The truth is, that journey has been a significant point of friction. The velocity of change we’re seeing now with AI, whether it’s building agentic systems or simply a new chatbot with a large language model, is just too fast to wait for data perfection. The only wrong answer is to delay. The first step is to simply push forward and start making concrete decisions about building your AI platform of the future. Let that platform become the forcing factor. As you build it, it will naturally demand specific data, and that’s when you solve the integrity problems—not before. This approach flips the script; instead of letting imperfect data halt innovation, you let the drive for innovation force your data into shape. Your competitors are already taking advantage of this, so the urgency is real.
You highlighted the overwhelming complexity of having hundreds of choices across LLMs, infrastructure, and toolsets. Beyond a flexible provider, what specific metrics or frameworks should a company use to evaluate and select the right technology partners from this vast and complex landscape?
The landscape is incredibly daunting right now. You’re looking at four, five, even six different dimensions of choice—your LLM, your infrastructure provider, your development platform, your agentic AI framework—and in each of those dimensions, there are literally hundreds, if not thousands, of companies building solutions. The complexity is astronomical. So, rather than a rigid framework of specific metrics, I advise a shift in mindset. The most critical evaluation criterion is a partner’s ability to provide maximum flexibility and lower the barrier to entry. Ask them: how easily can I switch between different large language models? Can I scale my operations without having a massive in-house team to physically stack GPUs and deploy software? The key metric becomes agility. You want a partner whose infrastructure allows you to make these critical choices dynamically, without getting locked into a single proprietary path, because the best tool today might not be the best tool a year from now.
Your example of an AI agent autonomously verifying a traveler’s purchases was compelling. Can you share another practical, real-world anecdote of how agentic AI can streamline a payment process and elaborate on the main technical hurdles to integrating it with a bank’s general ledger?
Absolutely. The most transformative impact we’ll see is in fraud protection. Imagine an AI agent not just as a passive guard but as an active financial co-pilot. It sees payments going out and coming in, and when a suspicious transaction occurs, it does more than just send you an alert. It autonomously cross-references your calendar, location, and past behavior to validate it. The real magic—and the biggest hurdle—is tying this intelligent agent seamlessly into the bank’s core systems. The central question we must solve is how to take one of the 200 or 300 different agentic platforms focused on fraud and connect it directly to the general ledger or ERP solution. This integration is the critical challenge. It’s about creating a real-time, two-way conversation between a cutting-edge AI and a legacy system of record, enabling that instant fraud detection and verification without any human lag.
You predicted customers will switch to “AI-born” banks that support personal AI agents. What are the first three concrete steps a traditional financial institution should take today to begin fundamentally transforming its user experience and prepare for this new interaction model?
This is the most important thing for institutions to grasp: the very nature of user interaction is changing. First, leadership must fundamentally accept that the future isn’t another app; it’s about enabling a customer’s personal agent to perform actions on their behalf. The first step is a strategic and cultural commitment to this new paradigm. Second, they need to build the technical foundation. This means moving beyond monolithic systems and developing flexible, secure APIs that an authorized agent can interact with. You cannot support this future if your data and systems are locked away in inaccessible silos. Third, start small and prototype. Don’t try to build the entire system at once. Begin by allowing an agent to perform simple, non-critical actions. This allows the bank to learn and adapt, because if your institution doesn’t support this level of interaction, I guarantee customers will find an “AI-born” one that does.
What is your forecast for the evolution of data strategy in financial services over the next three to five years, especially as agentic AI becomes more mainstream?
Over the next three to five years, data strategy will undergo a radical transformation from a passive, archival exercise into the active, beating heart of the customer experience. The pursuit of the static “golden record” will be replaced by the demand for a dynamic, real-time “river of truth.” As customers begin to authorize personal AI agents to act on their behalf, a bank’s data infrastructure can no longer be a back-office function; it must be a front-line product. The winning strategies will prioritize immediate data accessibility and unimpeachable integrity, because an AI agent making autonomous financial decisions requires a level of trust and speed that today’s batch-processed systems simply cannot provide. Data strategy and user experience strategy will become one and the same.
