Financial services firms are moving quickly on AI, but many are still treating it as a technology rollout rather than an operational redesign.
As banks, insurers and other financial institutions invest in AI agents, reasoning models and automation, the bigger question is how work, accountability and compliance need to change around them. Poorly planned deployments risk adding another layer to legacy systems without delivering the productivity, cost or customer-service gains firms expect.
Karli Kalpala, head of strategy & AI agent business at Digital Workforce, a business automation and technology solutions company, suggests that AI will only transform financial services if organisations rethink the way decisions are made, supervised and governed.
The financial services industry is discovering that AI success demands more than just adopting the technology. Despite significant investments in AI, 50 per cent of projects fail, not because the tools don’t work, but because organisations are deploying them without fundamentally rethinking their operations. If you want to be one of the winners in financial services, you have to think broader than adoption.
Back to basics
Countless technologies have been touted as the next industrial revolution, but AI has the best chance of living up to that claim. With every adoption of AI, we’re forced to rethink who or what does the work in an organisation going forward. What repeatable tasks is it adept at? What data can it analyse and make decisions on? Will this shrink the workforce?
AI is gaining traction in financial services because it is changing how information moves through organisations. Traditionally, a skilled worker would need to read a document, apply judgement, and produce another output – whether that meant assessing credit risk, processing an insurance claim, or completing a banking application.
The key characteristic of this work is that human judgement and reasoning is needed to map the unstructured data in the document to a structured decision, which can be then moved forward. And this is changing with AI as for the first time in human history we can create a new document from another document without using human brain as the inference unit. We see the possibility to deliver knowledge work as software, which will decouple human labour from business outcomes.
What can AI do for financial services?
When reasoning models and AI agents are applied to a sector that processes vast amounts of data every day, the impact becomes clear. In many cases, the only reason a person sat in the middle of the process was that a professional had to interpret the document and decide what should happen next. Agentic AI and the so-called reasoning models can now perform that inference, meaning resolution times and COGS for insurance claims, fraud investigations, and everyday customer processes – such as applying for a loan – will collapse.
At its most basic level, AI can support low-level operational tasks. At its most sophisticated level, it can drive decision-making alongside the human workforce, interpreting complex or unstructured data 24/7 and allowing teams to focus on work that requires deeper nuance and context.
The reality is that layering a technology on top of an existing legacy system and expecting transformation doesn’t address the changes that need to happen behind the scenes. Organisational structures need redesign, workforce education has to take place, and pilots that operate in isolation can’t deliver measurable business impact. Seeing real change and value requires reimagining roles, workflows, and decision-making frameworks.
Instead of starting with the as-is and applying continuous improvement frameworks the real opportunity with AI is to build an AI native core and migrate legacy operations over time. Key question to address is; what decisions needs to absolutely stay with a human due to regulatory or business controls, and by definition everything else can be automated in the new AI native core.
Why compliance could be AI’s greatest foil
Given that more than 75 per cent of UK financial services are now using AI, there’s a serious gap emerging between what organisations expect to gain from AI and what is materialising. Crucially, compliance is being treated as an afterthought, an uncomfortable thought for a sector that hinges entirely upon it.
Even more concerning is the ‘wait and see’ state that many financial services find themselves in. In doing so, they risk both a diminished ROI, as well as the protection of consumers and businesses who use their services, should a major incident occur.
The greatest differentiator for financial services is accepting that, even with AI in a form of regulatory limbo, compliance cannot be an add-on. AI and automation can speed up processes and reduce reliance on legacy systems, but with the EU AI Act and EIOPA quickly tightening oversight, financial services innovating without compliance in mind will find themselves exposed, and no amount of technology can undo reputational damage.
To bring this into practice, any autonomous AI solution under the EU AI Act needs continuous monitoring and audit trail against ‘model drift’, the fact that the underlying LLM changes behaviour over time. Not to mention that the API endpoints and just natural evolution in model performance need continuous attention from the operations teams. It is not too far-fetched to state that 80% of the costs are in the RUN phase and very few organisations have thought about this when they make build vs. buy decisions.
How to avoid the upgrade trap?
The financial services sector must look past claims that adoption = progress, and treat AI as an operational overhaul as much as a technical upgrade. Forging a path forward means addressing structural barriers that prevent AI from delivering bottom-line results, including, unclear accountability, and failure to prepare workforces to supervise AI effectively and compliantly. For example an insurer don’t need a chief claims officer who owns the claims management process, but a chief ai accountability officer who owns the outcomes of the AI-native claims operation.
Rather than focusing on a large-scale rollout of AI to thousands of employees in the hope that productivity will follow, the mantra for this sector should be that the future of financial services is a handful of people overseeing hundreds of AI agents. Workforces will evolve, jobs will change, and resistance to this is natural. The organisations that can accept this and equip human workforces with the education and clarity needed to collaborate with AI will see real value. The technology is there, and it’s more than a capability sitting within enterprise and IT architectures.
Addressing the AI blindspot
In hot pursuit of AI, and working to ensure the sector doesn’t fulfil its reputation as lagging behind, the financial services sector risks missing some of the biggest blind spots. The greatest risk to ROI is assuming that adopting this infinitely scalable reasoning tool is purely a technological challenge. The real work will require organisational change, and financial services will quickly limit themselves if they reduce AI to a software development exercise.
The post Addressing the AI Blind Spot: Why Tech Alone Won’t Transform Financial Services appeared first on The Fintech Times.
