Andre Nedelcoux, Global VP Technology, FSI at Intellias
The narrative around artificial intelligence (AI) in financial services (FS) has largely been dominated by a widening gap.
Across industries a small group of organisations is already pulling sharply ahead. According to PwC’s global AI Performance study, nearly three-quarters (74%) of AI’s economic value is captured by just one-fifth (20%) of companies, highlighting how quickly advantage has concentrated among a select group of leaders.
In FS, that dynamic is playing out in real time. Large, well-capitalised institutions are accelerating ahead, deploying AI across front, middle, and back-office functions, realising measurable return on investment (ROI), and embedding capabilities at scale.
A small group of frontrunners accounts for the majority of scaled deployments and realised value, and that matters because advantage compounds. While these early movers attract top-tier talent, strengthen data pipelines, and build proprietary capabilities, many mid-tier firms are watching on, paralysed by inaction, constrained by legacy technology, fragmented data estates, and increasing board-level scrutiny over investment returns, ultimately convinced they’re already too far behind to compete.
But that’s a premature – and potentially dangerous – conclusion to draw. The real risk facing FS firms today is not falling behind. It’s failing to act while the window of opportunity remains open. And let me be clear – it is still open.
Why speed now favours the fast followers
The reality is the current phase of AI adoption could favour those who have not yet fully committed – provided they act decisively now.
How? Because the economics of AI delivery have fundamentally changed. What previously required multi-year transformation programmes – often involving wholesale system replacement – can now be achieved in a matter of months using modern tooling.
Cloud-native AI platforms have reduced infrastructure barriers, while pre-trained models and application programming interfaces (APIs) are accelerating development cycles. At the same time, low-code and no-code environments are enabling faster prototyping, and increasingly mature vendor networks are providing off-the-shelf accelerators for common FS use cases.
This creates a unique dynamic: mid-tier firms can “fast follow” by learning from the experimentation of early adopters, avoiding sunk costs, selecting proven use cases, and implementing with greater precision. The opportunity now is to apply those lessons at speed.
The real barrier is not technology, it’s paralysis
Despite these advantages, many organisations remain stuck. Three structural barriers consistently emerge:
1. Legacy technology and fragmented data: Decades-old core systems and siloed data architectures make integration complex. But waiting for full modernisation is no longer viable, or necessary. Organisations can still unlock value from AI before their data challenges are fully resolved, and AI itself can also help address and streamline those data challenges.
2. ROI scrutiny at board level: Executives are under pressure to justify AI investment with clear, short-term returns. This often leads to over-analysis and under-delivery. This is also a cultural challenge: because of the pace of innovation around AI, building multi-year business cases is less relevant and it’s about testing various hypotheses and iterating over AI use cases as the technology matures.
3. Organisational risk aversion: In a highly regulated sector, caution is understandable. However, excessive caution can become a strategic liability. There is also a strong alignment between building AI systems that are explainable to regulators and creating the feedback loops needed to improve performance over time – what satisfies the regulatory scrutiny also helps improve ROI.
4. Fear of change: As applications of AI become more advanced, there is genuine fear of job replacement across the whole organisation. Change management and leadership focus is required to address this new way of working where AI augments and multiplies what humans do.
The path forward is depth over breadth
For firms looking to act, the way forward is not to attempt enterprise-wide transformation overnight. Instead, the most effective strategy is focus, because depth over breadth will always come out on top.
Leading fast-followers tend to concentrate investment in three or four high-impact, well-understood use cases with specific sub-verticals. In online investment and wealth management, this includes instant and continuous Know Your Customer (KYC) processes, client investment copilots, and compliance oversight agents. In asset management, we have seen success with continuous investment research, alternative data processing, and client servicing agents.
By focusing on targeted areas, firms can deliver measurable ROI within months rather than years, build internal confidence and stakeholder buy-in, and develop reusable capabilities that can be scaled over time. At the same time, rapid advances in agentic AI development frameworks – such as the way Anthropic is reshaping aspects of the software development lifecycle – mean that small expert teams can now build prototypes and put them in users’ hands within weeks.
Crucially, this approach also enables the parallel development of AI-ready data foundations, which are central to scaling effectively: improving data quality, governance, and accessibility without requiring wholesale transformation upfront. The mindset needs to shift towards working backwards from value for clients and the organisation, then iteratively building the foundations needed to support and scale these AI applications.
Building capability without creating dependency
Another key consideration is how firms build capability. Relying entirely on external vendors may accelerate initial deployment, but it risks creating long-term dependency. Conversely, attempting to build everything in-house can slow progress dramatically.
The optimal approach for FS could be a hybrid one, to partner where it accelerates delivery, build internal capability where it creates strategic advantage, and ensure knowledge transfer is embedded from day one. Agentic AI development is a new skill engineering teams need to learn and it’s not just about writing code faster: the big challenge is rethinking the full development lifecycle from idea generation to production release, and applying it within the context of an existing set of enterprise systems. This requires specialist expertise, alongside strong change management to support adoption and integration across the organisation.
It’s a balance that allows firms to move quickly while retaining control over their long-term AI trajectory.
The window is still open, but it won’t always be
According to the European Banking Authority, digital transformation – including AI adoption – is now a key driver of competitiveness and resilience across the EU banking sector. Firms that delay risk not only falling behind peers but also struggling to meet evolving regulatory and customer expectations.
The narrative that “it’s too late” is both inaccurate and dangerous. It breeds inaction at precisely the moment when action is most critical. Modern AI tooling has levelled the playing field more than many realise. The barriers to entry are becoming lower, the timelines shorter, and the lessons from early adopters readily available. What matters now is not perfection, but momentum.
Firms that move decisively – focusing on high-impact use cases, building scalable data foundations, and developing internal capability – can still close the gap. But those that continue to stall on action may soon find that the opportunity to do so has passed.
The risk is no longer falling behind, it’s standing still while others move forward.
Andre Nedelcoux, Global VP Technology, FSI at Intellias
Andre Nedelcoux leads the strategic delivery of large-scale engineering programmes for major financial institutions across banking, insurance, and capital markets at Intellias. With two decades of experience at the intersection of technology and financial services, he specialises in helping organisations translate AI and data capability into measurable business outcomes.
Andre’s career spans senior leadership positions at AWS, Mesh-AI, and Luxoft, where he built and scaled technology practices serving some of the world’s most complex regulated enterprises. He brings particular depth in cloud architecture, data engineering, and AI adoption — disciplines he has applied across both global tier-one institutions and high-growth challengers navigating digital transformation.