AI leadership matters

A thought-leadership perspective on how AI may reshape roles, leadership and operational discipline in banking. Many banks still talk about artificial intelligence as if it were simply the next digitisation programme. In our view, it is more than that. Digitisation moved processes from paper to screens. AI can automate parts of analysis, pattern recognition and decision support. For banking, this may represent a broader change than a purely technical upgrade. The pressure is likely to become visible first in parts of the bank where data, processes and operational discipline matter most: trading, custody, operations, client servicing, risk management, legal and compliance, marketing, reporting and reconciliation. In these areas, AI can help reveal which activities create measurable value — and which activities mainly compensate for fragmented processes, manual workarounds or poor data quality. 

In trading, AI can support the identification of patterns in client behaviour, help flag anomalies and provide suggestions before a human has reviewed all available data manually. In custody, the situation is similar. Corporate actions, settlement issues, fee analysis, reconciliation, regulatory checks and client reporting are important activities. Many of them are data-heavy and error-prone, which makes them relevant areas for AI-supported improvement.

In the near term, the activities most exposed to change are likely to be those furthest away from judgement: manual data entry, routine checks, standard reporting, pure information gathering, first-level clarifications, manual handovers between systems and spreadsheet work without meaningful analysis behind it. Many jobs may not disappear in formal terms. However, the substance of certain roles may change significantly. If a large part of a role can be automated or supported by AI, the person in that role will need to develop new skills and redefine how they create value. That is the real wake-up call. Those who do not learn may gradually become less relevant in their current form of work.

Experience remains highly valuable — but only if it stays connected to new tools and data-driven ways of working. A trader, custody specialist or operations professional with many years of experience has knowledge that no system can simply invent. If that knowledge is not linked to data, platforms and modern workflows, it remains valuable but difficult to scale across an organisation.

Experienced professionals may therefore be among the biggest beneficiaries of AI — provided they are willing to connect expertise with technology. AI needs context. It needs people who can judge whether a result is plausible, whether a market is behaving unusually or whether a settlement issue that looks harmless could later become costly. That judgement remains highly valuable. But knowledge that never leaves the individual expert has limited organisational impact.

For younger professionals, the opportunity is also significant. They may no longer need to spend years on administrative routines before they are exposed to more complex business questions. With AI, they can learn faster how clients, markets, risks, processes and profitability are connected. But the same principle applies: using AI only to produce faster, more superficial answers does not create a sustainable advantage. The real edge comes from learning faster and thinking more clearly.

The key skill of the future may not be operating AI itself. That will likely become table stakes. The more important skill is making better decisions with AI. Using tools is not, on its own, a capability. Capability begins when people understand which questions matter, which data matters, where the limits are and what a decision actually means in the real business context.

In banking, that raises an important question: who understands market structure, liquidity, custody costs, settlement risks, regulatory requirements, client profitability and operational dependencies well enough to steer AI responsibly? The competitive advantage is not the prompt alone. It is the context behind it.

Performance measurement may also change. Today, activity is still measured too often as a proxy for contribution. Over time, that logic is likely to weaken. What may matter more is:

• Who reduces errors?
• Who automates or improves processes?
• Who contributes to better margins?
• Who identifies risks earlier?
• Who helps reduce costs in trading or custody?
• Who enables others to work faster or better?

The central question is likely to shift from "What have you done?" to "What has improved because of you?" — with increasing focus on measurable impact over the next 12 to 24 months.

That shift may make some people uncomfortable. It can expose apparent efficiency, roles that mainly exist because processes are broken and organisations where information is protected rather than shared to create value.

AI may also change leadership. Spans of control could widen as simpler management tasks become more automated or more transparent. What remains are the more difficult parts of leadership: setting direction, making decisions, creating motivation, resolving conflict and leading people through uncertainty. Leaders who see AI only as a cost-cutting tool may miss the larger opportunity.

Leaders who ignore it may also fall behind. Leadership increasingly means organising learning — not as optional training at the edge of the calendar, but as a practical and regular part of work.

When someone says, "My people do not have time for AI," it may also mean that learning has not yet been given the right priority. If a bank is serious about AI, learning should be treated as part of the work itself: structured, practical and linked to real use cases from daily operations.

The ethical question is not only whether AI is used. It is how responsibly people are led through the change. It would be wrong to leave employees alone with this transition. But it would also be misleading to suggest that nothing meaningful will change.

In operations, the pressure may be especially strong. Margins are tight, complexity is rising and clients expect more transparency. Manual work is becoming increasingly difficult to justify where better, safer and more scalable alternatives are available.

Those who do not engage with AI in these areas may not only miss efficiency gains. Over time, they may also lose relevance in the way value is created and measured.