Vernon Yai stands at the intersection of data integrity and the rapid evolution of financial technology, bringing years of expertise in safeguarding sensitive information within the banking sector. As a thought leader specializing in data governance and risk management, he has observed firsthand how the world’s largest financial institutions are transitioning from AI experimentation to deep operational integration. This conversation explores the massive scale of AI adoption at firms like Bank of America and Citigroup, the shift toward agentic workflows, and the industry-wide debate over whether the true value of artificial intelligence lies in corporate margins or the customer experience.
With hundreds of thousands of employees generating nearly half a million AI prompts daily at institutions like Bank of America, how do you manage the operational shift while ensuring such a massive scale remains productive?
Managing an ecosystem where over 200,000 employees are interacting with AI-enabled capabilities requires a rigorous governance framework that balances innovation with security. When you see a workforce generating more than 400,000 prompts every single day, it highlights a fundamental shift in how tasks are prioritized—moving away from manual entry toward advanced agentic workflows. We are seeing banks move beyond simple experimentation to 34 fully implemented use cases that handle everything from coding support to intensive meeting preparation. The key is ensuring that these hundreds of thousands of daily interactions don’t just create digital noise but actually drive the consistency and client service that leaders like Brian Moynihan are reporting to their investors.
Can you elaborate on the specific ways these AI use cases are transforming the day-to-day lives of financial advisors and developers within these massive organizations?
The transformation is particularly visible in wealth management, where AI-powered tools are now pulling directly from Salesforce CRM data to help advisors prepare for client meetings with a level of depth that was previously too time-consuming. At Bank of America, they have identified over 300 approved use cases, including 114 specifically for generative AI, which are designed to strip away the heavy manual work that CFO Alastair Borthwick identified as a major bottleneck. For developers, coding support tools are becoming a core part of the infrastructure, allowing them to build and deploy software with much higher efficiency and speed. By embedding these tools into risk, finance, and technology sectors, banks are creating an environment where a teammate’s productivity is no longer capped by administrative burdens.
Citigroup has reported that nearly 90% of its workforce is now utilizing AI tools; what does this high level of adoption tell us about the future of human-AI collaboration in banking?
When 9 out of 10 people in an organization as massive as Citi are actively using these tools, it signals that AI is no longer a niche project but a baseline requirement for modern banking operations. CEO Jane Fraser has noted that this widespread adoption is accelerating the speed at which products are brought to market, which is a massive competitive advantage in a crowded field. It is not just about giving people tools; it’s about a large-scale technology implementation strategy, evidenced by bringing on new leadership like CIO Brian Saluzzo to scale these functions across the entire enterprise. This level of engagement suggests that the “transformation work” is maturing, allowing employees to focus on high-value growth and complex client experiences rather than routine processing.
There seems to be a divergence in opinion regarding who actually captures the value of AI, with some CEOs focusing on long-term shareholder value and others, like Jamie Dimon, suggesting the customer is the ultimate winner. Where do you see the balance?
It is a fascinating tension because while BNY’s Robin Vince sees AI as a significant source of long-term value for employees and shareholders alike, Jamie Dimon is very candid about the high costs involved. JPMorgan has nearly 1,000 live AI use cases spanning risk, fraud, and marketing, yet Dimon is quick to point out that these investments are expensive and might not boost company margins in the near term. From my perspective, the immediate benefits are operational—things like document reading and fraud detection—but the competitive nature of banking means those efficiencies eventually get passed down to the user. As usage scales, the banks that can navigate the high expense of AI while maintaining client-centric growth will be the ones that survive the mounting margin pressure.
What is your forecast for the role of AI in the financial sector over the next two years?
I expect we will see a shift from broad experimentation to a “quality over quantity” phase where the 1,000 or more use cases we see today are distilled into incredibly powerful, seamless agentic systems. We will likely see more specialized executive roles focused entirely on scaling these technologies safely and efficiently while maintaining strict data governance. While the initial costs are heavy and may not immediately reflect in company margins, the long-term payoff will be a banking environment where fraud is caught in milliseconds and financial advice is hyper-personalized for every single customer. Ultimately, the next 24 months will be about proving that those 400,000 daily prompts translate into measurable financial stability and a significantly better experience for the people these banks serve.
