Attributed to Jacob Rider, UK Payments Lead Projective Group
The financial services industry has rarely invested so heavily in a single idea. AI, and more recently generative and agentic AI, now sits at the centre of boardroom agendas, capital allocation and competitive positioning. Yet for all the investment, the value is far from straightforward. Some players, particularly those that began building AI capabilities before the GenAI wave, seem to be pulling ahead, while others are still some way from converting experimentation into measurable gains.
This is not unusual. History suggests that transformative technologies rarely deliver immediate results. Electricity took decades to reshape manufacturing. The internet followed a similar pattern. In each case, early adoption focused on layering new tools onto old processes. The real gains only arrived when organisations rethought how work was structured.
Financial services is at a similar moment where AI is everywhere but its impact, while rapid, remains uneven. The challenge now is how institutions are prepared to change around it.
The illusion of progress versus where change is starting to show
Much of today’s AI activity in banking still sits in what might be called an optimisation phase, with the biggest gains being seen at a productivity level. Processes are faster, outputs are cleaner and costs are lower in certain areas. Risk and compliance functions, for example, are using AI to process data at a scale no human team could match. Fraud detection is becoming more proactive and regulatory interpretation is quicker.
These are immediate and meaningful improvements but they do not fundamentally alter how banks operate. The underlying workflows remain intact and therefore, AI is being used to sharpen existing ways of working, rather than redesign the entire model.
This certainly helps explain the gap between expectations and reality. When organisations invest in new technology but keep legacy approaches intact, productivity gains tend to be incremental. The benefits are real but can feel limited. There are, however, areas where something more structural is beginning to emerge.
In retail banking, the shift towards AI-driven interaction is changing the shape of the customer relationship. Traditionally, the bank owns the customer relationship through
branches and digital channels. Increasingly, that relationship is mediated by intelligent systems that guide decisions, automate transactions and operate continuously in the background. Banks are now facing the challenge of how to redesign the customer relationship and decide where they want to differentiate from other banks: in the strength of the intelligent system, or in the human interaction layered on top.
Investment banking offers a different perspective. Here, AI is starting to erode the traditional pyramid model. Tasks that once justified large teams of junior analysts: synthesising data, preparing materials, supporting deal execution can now be handled far more efficiently by machines. Expertise is still essential but its centre of gravity is shifting. Value increasingly sits with those who can interpret, challenge and make decisions under uncertainty and with firms developing proprietary models on their own data instead of relying purely on commoditised tools. The structure of the firm, built around layers of manual work, becomes harder to justify.
The real shift: from knowledge to judgment and why large organisations struggle
At the heart of this transition is a more subtle change in how value is created. For decades, financial services have been built on the control and distribution of knowledge. Information asymmetry was a source of advantage. Expertise was tied to access – to data, to models, to experience accumulated over time.
AI is compressing that advantage. Access to information is no longer scarce and analysis can be generated instantly. The middle layer of the value chain e.g. the gathering, processing, and presenting knowledge, is becoming commoditised. What becomes more important is what sits around it. While AI models are extremely good at answering questions; they still struggle with defining the right questions to ask. In other words: curiosity, curation and judgment.
These are not new skills, but they are becoming more central. As machines take on more of the analytical workload, the human role moves closer to defining problems and taking responsibility for outcomes.
Traditional financial institutions were designed for a different era. Their structures reflect a world in which scale, control, and standardisation were the primary sources of advantage. Layers of management, complex processes and extensive systems all served that model. AI does not fit neatly into this environment. It requires speed, adaptability and integration across functions.
For new digital players, more agile firms, this creates an opportunity. They can build AI native infrastructures from the outset, embedding AI into IT systems rather than retrofitting it
onto them. They can operate with fewer layers and less overhead. For incumbents, the challenge is more complex. The existing model continues to function but it becomes increasingly inefficient. The cost of maintaining it rises, while the benefits of change are difficult to capture without disrupting the organisation itself.
A turning point, not an endpoint
It is tempting to see the current moment as underwhelming. The investment is visible but thus far, the impact is not. But this is how these transitions tend to unfold. The infrastructure is built first. The organisational change follows more slowly and the impact, when it comes, is how it reshapes the system around it.
Financial services is approaching an inflection point but at very different speeds. Some institutions are moving into more fundamental change, while others are still seeing incremental gains, particularly where control in the value chain is shifting.
The institutions that navigate this successfully will be those that treat AI as a catalyst to rethink existing processes. That requires uncomfortable decisions about structure, roles and where value truly lies. The technology is ready. The question is whether the industry is.
Image provided by Lead Projective Group
