Mario Ernst analyzes the structural mistakes that are leading financial institutions to accumulate proof-of-concept initiatives, isolated automations, and limited real impact from AI. He explains what it takes to move from pilots to true transformation.
Artificial intelligence reached the financial industry at a pace that most financial institutions found impossible to anticipate. Almost as a matter of corporate survival instinct, adoption efforts moved quickly across the sector, often without clear governance frameworks or strategic planning.
Implementations multiplied, but integration and real transformation, not so much.
For Mario Ernst, CEO and founder of Evolution Labs, the problem doesn’t lie in the technology itself, but in how organizations decide to adopt it and capture value from it.
“There’s a widespread belief that organizations have to adopt AI, but we’re making the wrong decisions about ownership and resources. That’s leading to innovation operating models that often fail to impact the organization’s financial results. In a way, it becomes a self-fulfilling prophecy for innovation skeptics,” the specialist emphasizes.
When organizations don’t truly believe in the initiative and move forward mainly due to competitive pressure, many fail to manage it with the level of design, governance, and operational discipline it requires.
“As a result, AI-driven innovation doesn’t deliver concrete value, and organizations end up reverting to traditional approaches,” says Ernst, interviewed by Abstracta.
Behind this pattern, he points out, there are unclear decisions around leadership, methodology, and innovation budgets:
“Many executives blame AI innovation for not delivering on its promise of value creation, but they fail to see that what’s really happening is that they’re not implementing it correctly.”
The scene repeats itself across many executive committees: enterprise licenses, teams experimenting with agents in an unstructured way, pilots running in different areas, and an implicit expectation that, through accumulation, “something will eventually scale.”
However, when different organizations approach Abstracta in their search for AI-driven transformation and we ask how they have integrated AI, which critical business processes have changed, or how they are controlling the new operational risk, the answer is often vague.
The Blind Spot of AI-Driven Transformation
Gradually, many organizations are beginning to realize that the sum of individual initiatives doesn’t amount to transformation. For Mario Ernst, the core problem lies in adoption efforts that move forward without a shared map of how the entire business operates.
His diagnosis comes down to a phrase that invites self-criticism and deeper analysis: “there’s no end-to-end process perspective.” Ernst argues that when adoption becomes fragmented, organizations lose two things at the same time: the ability to govern risk and the ability to capture value. The result is progress, but not necessarily building something sustainable.
The Symptom: Many Pilots, Little Transformation
In day-to-day operations, this lack of an end-to-end perspective shows up as local efficiency: each area improves its own segment, automates tasks, speeds up decisions, and may even reduce internal friction.
The problem emerges when those segments aren’t connected to the full flow of a financial operation, which typically spans business, risk, compliance, technology, operations, and channels. “Each area looks only at its own functional silo, which leads not only to a loss of process visibility, but also to use cases that fail to monetize.”
Once this logic takes hold, the initiative portfolio grows, but it becomes difficult to answer basic questions: which critical processes are being redesigned end to end, which controls are being added, which risks are being accepted, and what economic impact has actually been validated.
“We see organizations full of proofs of concept (POCs), minimum viable products (MVPs), and testing efforts, but lacking projects that are truly transformational and that are capturing, creating, and capturing value,” the expert emphasizes.
The Root Cause: Adopting Technology Before Designing the Operating Model
Ernst observes that many institutions begin their AI adoption by broadly rolling out tools across the organization, hoping that innovation will emerge from usage.
For him, that sequence leads to two predictable effects:
- Redundant or incompatible initiatives: different automations addressing the same problem, or local improvements that create new friction at another stage of the flow.
- A lack of shared criteria for prioritizing use cases with cross-cutting impact.
Mario Ernst in the First Person,
CEO & Founder at Evolution Labs
Why are there so many use cases that never get monetized?
Because organizations optimize isolated segments and lose sight of the end-to-end process, which is where control and value actually materialize in the financial industry. A single function may become more productive, but if the overall flow doesn’t change, the business doesn’t capture the benefit in costs, revenue, conversion, or risk mitigation.
Why do you think so many AI pilots get scaled without being truly ready?
In too many cases, when someone asks why a pilot is being scaled, there’s no quantitative backing. Teams lack solid metrics and move forward simply because the project has already been tested or built. They jump straight into execution, as if it were a traditional project. That’s where the problem lies: the first phase of any project should always focus on learning, not monetization.
What should a project have in order to move from pilot to scale?
First, it needs clear governance. Second, it needs a strategy that connects experimentation with scaling—the so-called AI Blueprint. At its core, this is the plan that allows organizations to start with small use cases and POCs and end up with scaled projects that cut across the organization. That’s where monetization happens.
The Risk That Grows Quietly
Beyond returns, there’s a second consequence that often gets underestimated: operational risk. AI redesigns processes, automates decisions, and accelerates flows, but it doesn’t always leave a clear record of what changed, what was delegated to models or agents, and how it’s being controlled.
Ernst describes this as a recurring pattern: “organizations don’t always identify the risks of these new technological and automated processes, and therefore don’t put the right controls in place.”
In regulated institutions, that gap quickly turns into friction: audits, compliance issues, internal disputes over ownership, and a loss of end-to-end visibility. When that happens, the initial momentum behind adoption becomes defensive.
The Opportunity: A Moment Where “Everything Is Still to Be Done”
Ernst interprets the current landscape as a turning point for the sector: a shift from an industry focused on executing what is already defined to one forced to create, adapt, and differentiate itself. “There’s a lot of room to start creating things that didn’t exist before and to stand out,” he says.
For those leading innovation, technology, and operations, the opportunity lies in moving away from adoption by accumulation and toward adoption by design: end-to-end processes, clear priorities, shared metrics, and governance that allows organizations to scale without losing control.
“This is a moment where we have everything still to do,” he concludes.
How Abstracta Can Help in This Scenario
At Abstracta, we work with banks, fintechs, and financial organizations at both early stages of AI adoption and more advanced phases of transformation.
With nearly 2 decades of experience and a global presence, Abstracta is a technology company that helps organizations deliver high-quality software faster by combining AI-powered quality engineering with deep human expertise.
Our approach is built around three pillars:
End-to-End Visibility of Critical Processes
We support organizations that need to organize and redesign flows already impacted by AI, as well as those looking to identify, from scratch, where it makes sense to apply artificial intelligence across complete business processes.
We start by reviewing the organization’s current delivery process end to end (how software is built, tested, released, and operated) to identify bottlenecks, quality gaps, and structural constraints before introducing AI.
Governance and Responsible AI Deployment
We help define governance models, prioritization criteria, and maturity metrics to turn proofs of concept into production-ready capabilities—or to build that path in a structured way from the outset.
As part of this foundation, we design dashboards with relevant delivery and engineering KPIs (including DORA metrics) to establish a clear baseline and objectively measure the impact of AI adoption over time.
Quality Engineering, Risk, and Security by Design
We integrate advanced testing, observability, controls, and continuous validation into AI initiatives. This approach reduces operational and regulatory risk in both existing projects and new implementations.
Based on a solid quality strategy, we identify where AI can generate the highest impact across the delivery lifecycle and define how that impact will be measured. We complement this with structured training programs for the entire team to build sustainable internal AI capabilities.
We believe that actively bonding ties propels us further and helps us enhance our clients’ software. That’s why we’ve built robust partnerships with industry leaders, Microsoft, Datadog, Tricentis, Perforce BlazeMeter, Saucelabs, and PractiTest, to provide the latest in cutting-edge technology.
If you’re looking for a partner for software development in the financial sector, we invite you to explore our solutions and contact us.
Follow us on Linkedin & X to be part of our community!
Recommended for You
Introducing Abstracta Intelligence: Our New AI-Powered Delivery Platform
Fintech Expansion Checklist for New Countries
AI SaaS Product Classification Criteria for Banking and Fintech
Abstracta
