Across financial services, a provocative idea is gaining traction: the AI-run bank. In this vision, AI agents will make lending decisions, reconcile accounts, monitor compliance, and interact with customers. There will be human oversight, but over a largely autonomous system.
The appeal is clear: much greater efficiency and scalability than has ever existed before in banking. But framing this future as an “AI-run bank” misses the real challenge. The question facing leaders is not whether AI can perform banking tasks — it can — but whether banks are prepared to redesign work, leadership, and governance to make a truly AI-enabled organization a reality. It requires a total reimagination.
The gap between AI investment and real organizational change
Today, two perspectives dominate boardroom discussions. Some executives envision banks operating with a fraction of today’s workforce, overseen by a small group of highly specialized experts. Others argue that trust, regulation, and systemic risk impose natural limits to AI deployment, ensuring that humans will remain central to banking operating models even as AI capabilities expand.
While executives broadly agree that AI will materially reshape banking, there is far less consensus — and urgency — around what this means for the organization itself. Data from the Oliver Wyman Forum highlights the disconnect: Only 54% of financial services CEOs express concern about whether their workforce is ready for AI, even as AI ambitions accelerate.
The gap is reflected in behavior. Nearly all institutions are increasing AI and technology investment, modernizing infrastructure, and experimenting across functions — yet organizational models remain largely unchanged. Decisions about skills, governance, accountability, and leadership are deferred, creating a mismatch between where investment flows and where the real barriers to AI adoption lie.
The winners will match their AI ambitions with equally bold system redesigns anchored in three major workforce shifts: from roles and tasks to skills and jobs, from managers of work to orchestrators of outcomes, and from overseeing headcount to governing hybrid human-AI capacity.
AI is shifting focus in banks from roles and jobs to skills and tasks
As AI matures, work is becoming more fluid. Static, title-based jobs give way to dynamic, skills-based contributions focused on problem ownership, cross-business execution, and effective human-machine collaboration. AI breaks the traditional link between roles and value creation.
This shift changes how work is allocated and how humans contribute. Rather than executing predefined tasks within a role, employees increasingly own problems, collaborate across silos, and work alongside AI systems that handle data-heavy and rules-based activity.
Some institutions are already moving in this direction. Standard Chartered launched a skills-based talent model in 2020 built around internal mobility and skills visibility. Today, more than 40% of employees use skills passports to access learning and project-based opportunities, according to the company. Mastercard introduced a similar AI-driven internal talent marketplace in 2022, combining opportunity matching with personalized career coaching.
Data from the Oliver Wyman Forum illustrates this shift. The fastest-growing capability needs in financial services are technology fluency, assembling interdisciplinary teams, and leadership — precisely the skills required in organizations where work crosses silos and outcomes define value. Yet few institutions systematically develop these capabilities, despite the positive impact it can have: The Oliver Wyman Forum reported that colleagues who feel like they are building skills are more likely to state that they enjoy their work (63% versus the average of 32% for financial services employees).
How AI is transforming work and skills in financial services
As work becomes more fluid and increasingly delivered through a combination of humans and AI, traditional management models start to break down. Leading through static roles, stable teams, and role supervision is no longer sufficient. What matters instead is the ability to drive outcomes across shifting combinations of skills, functions, and capabilities.
This gives rise to a new leadership role that is not prevalent in most banks today: the orchestrator. Orchestrators do not manage fixed teams or oversee day-to-day task execution. They coordinate end-to-end outcomes delivered through a mix of human expertise and AI agents, interpret signals from systems, and decide when human judgment is needed.
Specialist teams in this model are smaller but far more leveraged. AI absorbs much of the data-heavy and rules-based work, while human experts focus on judgment, exceptions, and decisions in which context and experience matter most. Performance is judged on outcomes delivered effectively and safely, not utilization. In the near term, this can help democratize access to specialized expertise: Fewer people are required to deliver the same impact, allowing smaller institutions to compete and innovate against larger incumbents.
AI is turning managers overseeing headcount into governors of hybrid capacity
These shifts will force a more fundamental organizational change. To realize the benefits of skills-based work, outcome-oriented leadership, and AI at scale, banks must rethink how their organizations are governed and set up. Performance can no longer be managed through people alone. Instead, banks must learn to govern hybrid capacity across humans and AI agents.
AI is increasingly fulfilling work that was previously done by humans — execution, coordination, documentation, and elements of decision-making. The assumption that work equals people, and capacity equals headcount, no longer holds.
This exposes the limits of today’s organizational design. In many banks, senior leaders operate with spans of control of roughly six to nine direct reports, particularly in functions such as technology, operations, and risk. These spans reflect the coordination and supervision requirements of human-only organizations. As execution and coordination are shared between humans and agents, those constraints change.
The implication is not simply flatter hierarchies, but different ones. Leaders can be accountable for broader scopes of work without adding layers focused on supervising information flow. What matters is clear accountability for end-to-end outcomes, with explicit decision rights and escalation paths in hybrid human-AI workflows.
This creates a new governance and planning challenge. Humans and AI agents both create capacity, cost, and risk. Banks will need hybrid governance and hybrid strategic workforce planning that integrates people, agents, and third parties into a single operating model. Institutions that continue to equate workforce planning with headcount will struggle to capture the full value of AI.
How this plays out over the next three to five years
Different choices around skills, leadership, and governance are already producing three distinct trajectories for banks implementing AI.
One is the false start. Some banks are moving quickly to automate high-stakes workflows before skills, leadership models, and accountability are in place. As they progress, execution will outpace organizational readiness, leading to incidents that force rollbacks, while costs rise and trust erodes.
Another trajectory is disintermediation. Some banks are moving too slowly, treating AI as a constrained productivity tool and deferring deeper organizational change. As they wait for clarity, competitors and platforms will build new hybrid human-AI operating models that deliver superior customer experiences, progressively disintermediating incumbent banks.
The third scenario is disciplined reinvention. A subset of banks will invest intensively and intelligently in core capabilities — particularly data, platforms, and organizational redesign — to build optionality for an uncertain future. Rather than blind experimentation, they will focus investment on the foundations needed to support a hybrid human-AI operating model. These institutions will use their trust advantage to adapt as the future becomes clearer, scaling where it matters and pulling back where it does not.
One mitigating factor across all scenarios is regulation. In some markets, regulation will slow adoption for everyone. Progress will be constrained, insulating banks from both failure and disintermediation — but also limiting the benefits of disciplined agility. Change will continue, but at a more measured pace, shaped primarily by external constraints rather than internal choices.
A future of AI agents and humans working together in banking
AI will not run the bank — but it will run far more of the bank’s operational and analytical machinery than today’s organizations are designed to govern. That is the real leadership challenge.
The uncomfortable truth is that most institutions still treat AI as a tooling program, when it is in fact an operating model rewrite. To succeed, leaders must move beyond a human-versus-technology debate and be explicit about where human judgment is required in each workflow — and who is accountable when humans and AI agents jointly produce outcomes.
Credibility with regulators, investors, and employees will hinge on execution. Leadership teams will need to manage AI-generated productivity like any other form of capacity, with clear accountability, prioritization, and financial discipline — first implementing AI effectively and then converting the capacity it creates into measurable bottom-line impact.
The banks of tomorrow will use hybrid, outcome-oriented operating models. Getting there requires a shift as fundamental as any regulatory change could demand.
This article is part of our Known Unknowns report, highlighting the debates that will shape the future of financial services in the age of AI.