One of the most consistent observations from large-scale AI programmes is the persistent gap between what the technology could do and what organisations are actually doing with it.
In Part 2 of our CDO Magazine conversation, between Karan Jain, CEO at NayaOne and Sanjay Sankolli, Head of Data and AI Architecture and Intelligent Automation at Truist, they moved from the structural reasons why so many initiatives stall to the areas where AI is already creating measurable – if still modest – impact inside large financial institutions.
The discussion stayed firmly grounded in current reality rather than future speculation. What emerged clearly from our exchange is that today’s wins are overwhelmingly about augmentation and workflow acceleration, not autonomous decision-making.
Patterns of Value Across the Value Chain
The examples explored follow a consistent pattern across banking and insurance:
- In the front office, AI supports customer-service deflection, predictive servicing, retention efforts, and underwriting augmentation – providing better information or handling routine interactions so teams can focus on higher-value work.
- In the middle office, it refines fraud detection, accelerates KYC/AML triage, and supports claims processing through agentic workflows with strong human-in-the-loop validation.
- In the back office, document intelligence stands out: the ability to turn unstructured data into structured, decision-ready assets, often paired with intelligent process automation.
A particularly effective horizontal application is developer productivity tooling, which is delivering some of the clearest and most immediate enterprise-wide efficiency gains.
Efficiency First, Revenue Second
Most current deployments remain focused on bottom-line efficiency. Organisations are prioritising cost reduction and operational acceleration because these areas allow them to build trust and demonstrate value with lower regulatory and reputational risk.
Early signs of topline impact are appearing in select front-office areas, such as digital cross-selling and wealth management augmentation. Yet these revenue-oriented use cases are progressing more slowly, as they require higher levels of organisational trust, clearer regulatory guidance, and significantly stronger data foundations.
The Enduring Data Architecture Constraint
Even where value is being realised, the same foundational constraint repeatedly surfaces: the data and platform architectures of large financial institutions were shaped by decades of mergers, acquisitions, and project-by-project decisions. The result is fragmentation, islands of automation, and platform sprawl that limit the scope of what AI can achieve at scale.
Without meaningful data rationalisation, the wins remain real but bounded.
A Deliberate Pace
The broader pattern is clear: AI is gradually moving from back- and middle-office efficiency plays toward front-office value creation. The technology itself is no longer the limiting factor. The pace of adoption is determined by organisational, architectural, and data realities – and by the need to move deliberately in highly regulated environments.
Watch Part 1 here: Why AI Evaluations Fail to Scale
Watch Part 2 here: Where AI Is Delivering Value in Financial Services
The full four-part series examines the practical realities of moving AI from ambition to execution in regulated financial institutions.
We would value hearing which of these patterns align with – or differ from – what you are observing in your own organisation.
NayaOne provides the leading external AI evaluation platform trusted by major financial institutions to test, compare and de-risk third-party capabilities before any commitment. Book a guided walkthrough.
