When McKinsey publishes a major perspective on artificial intelligence (AI) in financial services, it tends to garner attention. Their recent piece, “Extracting Value from AI in Banking: Rewiring the Enterprise,” offers a thoughtful, structured, and compelling roadmap for banks aiming to scale AI enterprise-wide. For anyone involved in the commercial banking or technology strategy space, much of it rings true and aligns with what we at Datos Insights have been saying for some time.
That said, while the article articulates the “what” and “why” of enterprise AI well, it leaves the “how” somewhat idealized. Like many strategic frameworks, it describes the destination in crisp detail, but perhaps understates just how messy, political, and human the journey will be, especially within the complex organizational realities of most financial institutions.
In this blog post, I want to highlight a few areas where the vision is right, but the path is likely to be far more nuanced and complex than the article suggests.
AI’s Strategic Imperative Is Clear, but the Execution Gap Is Real
McKinsey correctly identifies the core challenge: banks must move from proofs of concept and isolated use cases toward holistic, AI-first transformations of core banking workflows. The emphasis on agentic AI and cross-enterprise capability stacks reflects the accelerating shift toward intelligent systems that not only generate insights but also take action.
Here’s the catch: Centralizing IT to enable business-led transformation is a tale as old as time. Banks and consultancies have been advocating for this model for at least the last two decades. In practice, the model frequently falters—not due to lack of vision, but due to the persistent inertia of organizational silos and internal politics.
Cross-functional alignment is rarely smooth, even in institutions with strong executive mandates. Business units protect their domains, and tech teams prioritize existing pipelines. Unless every layer of the organization is engaged, not just the C-suite, AI ambitions remain stuck in the “strategic wallpaper” phase.
Culture Eats AI for Breakfast
One of the more understated challenges in the article is internal buy-in. Banks are not monoliths; they are fragmented ecosystems of business lines, legacy platforms, and deeply entrenched cultures. Attitudes toward AI differ dramatically across roles and regions. The possibilities of agentic automation energize some business leaders, while others are openly skeptical, and often with good reason—from model opacity to compliance risk to sheer change fatigue.
For AI to scale meaningfully, banks will need to do more than define a top-down strategy. They’ll need to embark on a bottom-up journey of cultural change. Doing so includes educating front-line staff, equipping midlevel managers with the tools to lead change, and creating space for dialogue and dissent. In short, they must make AI transformation tangible to the thousands of people across the enterprise who are not sitting in AI steering committees but whose daily workflows and habits it will disrupt.
The human factors—reskilling, trust-building, and change management—will ultimately determine whether generative AI (GenAI) delivers operational value or becomes another shelfware technology. This is where many banks need more help.
Trust in AI Vendors Still Needs to Be Earned
McKinsey’s focus on agentic AI transforming core workflows is timely and necessary. At Datos Insights, we’re seeing growing interest in this model—AI agents that span business domains, automate multistep tasks, and integrate deeply with core systems. The promise is significant: fewer manual handoffs, faster decisions, and more consistent customer experiences.
However, the open question is: Who will banks trust to deliver this?
Much of this hinges on the vendor ecosystem. The reality is that many banks still operate on aging infrastructure with fragmented core systems. The idea of allowing an AI vendor, especially one offering large language model (LLM)-powered agents deep access to those systems, is, for many, still uncomfortable, if not an outright compliance impossibility. Trust in GenAI providers remains limited, particularly in regulated environments where explainability, auditability, and model risk management are table stakes.
Consider how long it took financial institutions to warm up to the public cloud. It wasn’t just a technical barrier; it was about trust, control, and regulation. AI will be no different. Some vendors (Temenos and Intellect, for example) are pushing aggressively toward more AI-native integration, but others (who will remain nameless here) are moving at a more cautious pace. Banks will proceed at different speeds, depending on their vendor stack, internal capabilities, and risk appetite.
A Clear Vision Isn’t a Short Distance
There’s a phrase we often return to at Datos Insights: “Don’t conflate a clear view with a short distance.”
The McKinsey article presents a crisp vision of the AI-powered bank of the future. It’s a vision with which we largely agree, but achieving it will not be a straight line. The journey will be uneven, full of competing incentives, legacy constraints, and human complexity.
That’s why we believe banks need more than a strategy—they need a path. That path includes realistic roadmaps, incremental wins, cross-functional alignment, and the ability to navigate internal resistance. It requires knowing not just what best-in-class looks like on paper, but what it takes to implement it in the real world—step by step, system by system, and team by team.
Where Datos Insights Comes in
This is precisely where Datos Insights can help. As an advisory firm focused exclusively on financial services, we understand the lived realities of the industry. We’ve worked with banks struggling to modernize onboarding, integrate legacy treasury platforms, and launch digital experiences that work for their clients. We’ve seen firsthand that technology is rarely the problem. The real bottlenecks are organizational: fractured incentives, risk-averse cultures, and lack of internal consensus on what AI is supposed to do.
A friend of mine used to manage large-scale modern enterprise resource planning implementations in global enterprises. The stories he tells are painfully familiar—teams clinging to fax-based processes, stakeholders resisting standardization, and change management plans never reaching the people doing the work. These weren’t failures of strategy or software. They were failures of human alignment.
The lesson for banks is this: AI transformation is not about plugging in the next tool. It’s about rewiring how the organization thinks, acts, and collaborates.
Final Thoughts: The Opportunity Is Real, but so Is the Resistance
AI is here to stay. In fact, two-thirds of treasury professionals in our latest research expect to be live with GenAI capabilities in their banks before the end of 2025. That’s a remarkable shift, but scaling that capability into something that drives real enterprise value will take more than LLMs and dashboards. It’s going to require banks to grapple with the hard stuff: power structures, workflow redesign, vendor fit, staff trust, and sustained cultural change.
McKinsey has provided a valuable map. But as any bank executive knows, maps don’t capture the terrain.
