AI is Not a Technology Problem

Every year, financial institutions collectively spend tens of billions of dollars on AI and data initiatives. And every year, a significant portion of that investment delivers far less than promised. The models are often sound. The data, increasingly, is available. The talent, while competitive, is accessible. So why do so many AI programs stall, underdeliver, or quietly get deprioritized after the pilot?

The answer, in my experience, is almost never technical. After nearly three decades in financial services, spanning finance, data, and AI leadership across some of North America's largest institutions, two US patents for AI innovations, and formal AI study at MIT and through the CAITL™ certification program, I have arrived at a conviction that shapes everything I now think about the future of this industry: we have been solving the wrong problem.

We have been treating AI as a technology deployment challenge. It is, in fact, an organizational transformation challenge, and the most consequential variable in that transformation is not the algorithm. It is the human being sitting in front of it, leading it, resisting it, or reimagining their work because of it.

Era of Business-Led AI Has Arrived — Ready or Not

For much of the past decade, AI in financial services was owned by technology. Data scientists drove the roadmap. Business leaders consumed AI outputs, often skeptically, rather than shaping AI strategy. That model is breaking down, and it needs to.

The next era will be defined by organizations where the business leads and technology enables. Where a CFO challenges a forecasting model intelligently, not just approves it. Where a Chief Risk Officer co-designs the governance framework for a credit model rather than simply signing off on it. Where frontline finance and risk leaders think in terms of AI-augmented judgment, not AI-delivered answers.

This shift is not optional. Generative AI has changed the calculus entirely. The tools available today, large language models capable of synthesizing regulatory filings, earnings reports, and internal management data into coherent, actionable narratives in minutes, are no longer the domain of data scientists alone. They are sitting on the desktops of analysts, finance managers, and risk officers right now. The question is no longer whether your organization will use AI. The question is whether your leaders are equipped to use it well, govern it responsibly, and build an organization that evolves with it.

Generative AI: The Inflection Point Finance Cannot Ignore

Generative AI represents a qualitative shift in what AI can do for financial services, not just in automation, but in reasoning, synthesis, and decision support. The traditional FP&A cycle, quarterly, backward-looking, consensus-driven, is structurally incompatible with the speed at which business conditions now move. Generative AI, combined with robust financial data infrastructure, makes continuous, forward-looking, scenario-rich planning not just possible but necessary.

Consider what this means in practice. A finance team that once spent three weeks preparing a board-level scenario analysis can now iterate in hours, not by removing human judgment, but by dramatically compressing the time between data and insight. A risk team can synthesize counterparty exposure across thousands of positions, regulatory signals, and market indicators simultaneously, surfacing the questions that matter rather than drowning in the data that doesn't. A CFO can ask a question of their financial data in natural language and receive a reasoned, sourced, auditable response, not a dashboard that requires an analyst to interpret.

These are not hypothetical futures. They are capabilities available today, and the institutions building the infrastructure, governance, and talent to deploy them at enterprise scale are accumulating a structural advantage that will be very difficult to close later. The window for proactive investment is open. It will not stay open indefinitely.

Foundation that Most Organizations Skip

The use cases that fail almost always fail for the same reasons, and they are rarely technical. Data that isn't trusted. Governance that exists on paper but not in practice. Business owners who were presented with AI rather than enrolled in it. And organizations that raced to production before building the cultural and operational infrastructure required to sustain what they deployed.

The foundational work, data quality, lineage, governance, architecture, is unglamorous. No one writes headlines about improving a data maturity score. But in every transformation I have been part of, that foundation is what determined whether the AI investment actually paid off. Cloud migration done right doesn't just reduce technical debt; it creates the unified, trusted data environment that makes AI possible at scale. A mature governance framework does not just satisfy regulators, it builds the organizational confidence that allows AI outputs to be acted upon rather than perpetually questioned.

The sequencing matters enormously, and getting it wrong is expensive. Organizations that skip the foundation in pursuit of visible use cases typically find themselves rebuilding it later, at higher cost, under greater urgency, and with the added burden of having to undo the credibility damage caused by AI systems that did not perform as promised. The leaders who insist on getting the foundation right first are not being cautious. They are being strategically intelligent.

Governance is not a Constraint — It is the Competitive Moat

There is a temptation, particularly during periods of rapid AI advancement, to treat governance as friction. I think that is exactly backwards, and it is one of the most costly misperceptions in the industry right now.

In financial services, model outputs influence credit decisions, capital allocation, liquidity planning, and regulatory reporting. The organizations with the most robust AI governance frameworks will move faster — not slower — than those without them. The reason is simple: trust is the rate-limiting factor in AI adoption. A CFO who trusts the forecasting model acts on it. A Chief Risk Officer who understands its assumptions is an advocate for deployment, not a gatekeeper. A regulator who sees evidence of disciplined model risk management is a partner in innovation rather than an obstacle to it.

The regulatory environment, SR 11-7, BCBS 239, the EU AI Act, and what will inevitably follow,  is not going to become less demanding as AI becomes more capable. Precisely the opposite. Organizations building governance capability now are building a structural advantage that compounds. The human-in-the-loop model is not a concession to regulatory caution. It is the right design principle for any AI system operating in a domain where the consequences of error are material. Not because machines cannot be right, but because organizational legitimacy, once lost, is very hard to rebuild.

People Equation: Engagement, Adaptation, and Honest Leadership

Employee engagement in the context of AI transformation is not a soft metric. It is a leading indicator of whether the transformation will hold. In my experience, organizations that achieve genuine AI adoption, not just deployment, but adoption, consistently share one characteristic: they actively invested in bringing their people into the transformation, not just communicating it to them.

That distinction matters. There is a significant difference between telling an organization that AI is coming and creating the conditions for people to engage with it, learn from it, experiment with it safely, and ultimately own a piece of it. Employees who participate in shaping how AI is used in their domain are fundamentally different from employees who are told how AI will be used. The former become advocates and innovators. The latter become sources of quiet resistance that accumulates over time into meaningful organizational drag.

But here is where leadership needs to be honest, genuinely, uncomfortably honest, about something the industry tends to soften: not every employee will make this transition. Some will engage enthusiastically. Some will engage with support and time. And some, despite goodwill on both sides, will find that the world AI is creating is not one in which their existing skills translate easily. Pretending otherwise is not kindness. It is a failure of leadership.

The right response is not to write those employees off. It is to think creatively and deliberately about how to redeploy their institutional knowledge, their relationship capital, and their domain expertise in roles where those assets matter, even if those roles look different than they did before. The organizations that handle this thoughtfully will retain institutional memory, maintain trust, and build cultures where people believe the organization will take care of them through change. Those that handle it poorly, or avoid the conversation entirely, will pay the price in engagement, attrition, and organizational cynicism at exactly the moment they need commitment.

What Forward-Looking Organizations Are Doing Differently

The financial institutions pulling ahead on AI share a set of characteristics that I believe will separate the leaders from the laggards over the next decade. They are worth naming plainly.

They have made AI fluency a leadership competency, not a technical specialty. Their senior business leaders understand AI at the level required to set strategy, ask hard questions, and govern responsibly,   without needing to understand the mathematics. They have invested in data infrastructure as a strategic asset, recognizing that the return on foundational data investment is not a single use case but the entire portfolio of AI capability that becomes possible once the foundation is solid.

They have built operating models that place genuine business ownership at the center of AI programs — not sponsors who approve budgets, but leaders who are accountable for outcomes and invested in the work. They have created structured mechanisms for employee participation, not engagement surveys, but actual forums, innovation challenges, and co-design processes that give people a stake in shaping the AI-enabled organization rather than simply receiving it.

And they have committed to the long arc of transformation with the same discipline they bring to financial planning: setting a clear direction, measuring progress rigorously, adjusting course without abandoning the destination, and sustaining leadership attention beyond the initial wave of enthusiasm. That last point is harder than it sounds. AI transformation fatigue is real, and the organizations that treat it as a sprint rather than a strategic capability will build will feel it.

Question Worth Asking

We are at a genuine inflection point. The AI capabilities available to financial institutions today, and the pace at which generative AI, in particular, is advancing, represent a once-in-a-generation opportunity to reimagine how finance works, how risk is managed, and how organizations make decisions under uncertainty.

The technology will keep advancing regardless of what any individual organization does. The models will get better. The costs will come down. The regulatory frameworks will mature. What will not happen automatically, what requires deliberate, courageous leadership, is the organizational transformation required to use these capabilities well.

Business-led AI is not a trend or a management philosophy. It is the only model that works at scale over time in organizations where the stakes are real and the consequences of failure are measured in capital, reputation, and trust. The institutions building toward it with intention and honesty, about the technology, about the governance required, and about the human journey involved, will define the next era of financial services.

The future belongs to the organizations that treat AI not as something that happens to them, but as something they are actively, thoughtfully, and honestly building. Together with their people.