Why banks risk repeating past QA mistakes

Michiel Willems
September 1, 2026

As banks accelerate investments in artificial intelligence, quality engineering and resilience testing teams are facing a familiar challenge: ensuring that ambitious technology transformations translate into real operational outcomes.

For financial institutions operating complex, mission-critical platforms, the adoption of AI does not simply introduce new tools.

It also creates new risks around governance, system reliability, testing coverage and operational resilience. Ensuring that AI-driven systems perform reliably, particularly in regulated environments, requires robust testing frameworks and strong foundations across data, architecture and organisational processes.

The issue is particularly relevant for QA and resilience engineering teams responsible for validating software behaviour before deployment.

AI-enabled platforms introduce new dependencies and operational complexity, meaning that banks must increasingly test not only functionality but also failure scenarios, system interactions and enterprise-wide alignment.

A recent analysis from consulting firm Capco, which is part of Wipro, suggests that many banks risk repeating familiar mistakes if they focus too heavily on AI tools without addressing deeper organisational and operational challenges.

In the analysis, which appeared in a Capco Intelligence article titled AI in financial services: new frontier, old habits?, the authors argued that financial institutions have historically invested heavily in new technologies but have often struggled to realise the full value of those investments.

“Financial services firms have never shied away from ambitious technology programs,” wrote authors Jessica Forbes, Chris Probert and Joe Forooghian.

The sector has seen multiple waves of transformation in recent decades, they noted. “From RPA to digital, cloud to data, the sector has repeatedly invested at scale,” the authors wrote.

Yet many of these initiatives have failed to fully deliver on their intended outcomes.

“Despite significant investment and effort, many of these transformations have not fully delivered on their strategic promise and ambitious investment returns.”

For QA and resilience engineering teams, these lessons are particularly important as AI becomes embedded into core banking platforms.

Foundations matter more than tools

According to the Capco team, the failure of transformation programmes is rarely caused by technology itself. “When projects fail to deliver it is rarely about a lack of effort, funding or technology,” Jessica Forbes wrote.

Instead, organisations often focus on highly visible aspects of transformation while neglecting the underlying organisational changes required to support them.

They argued that too much attention is often paid to “tools, pilots and performance targets” while not enough focus is placed on deeper organisational foundations such as “process change, cultural behaviors and governance.”

For banking technology teams, this insight has direct implications for how AI systems are tested and validated. Without clear governance structures, shared standards and aligned operating models, AI prototypes may work in isolation but fail to scale across enterprise systems.

Data and analytics investments

The Capco article also highlighted lessons from previous technology waves, particularly the adoption of big data and advanced analytics across financial services.

Many banks have invested heavily in data infrastructure but still struggle to translate those capabilities into measurable business outcomes.

“Financial services organisations have built vast data estates, many with strong architectural foundations,” the authors pointed out. However, they added, a persistent gap remains between technical capability and enterprise value.

“The gap between data capabilities and business outcomes still exists.”

One reason is fragmentation across teams and departments. “Teams pursued different priorities, built competing assets and measured success in inconsistent ways,” Forbes wrote.

Instead of improving outcomes, these investments often expanded the technology footprint without delivering enterprise-wide impact. “Investments expanded the technology footprint rather than improving outcomes.”

Moving beyond isolated pilots

The authors warn that similar patterns are already emerging in the current wave of AI experimentation. Banks are developing promising prototypes, but many of these initiatives remain siloed.

“The same pattern can be seen with AI, with teams developing promising prototypes without alignment on shared priorities, data standards or integration patterns,” they wrote.

This fragmentation creates major challenges when organisations attempt to scale AI solutions across complex banking environments.

“What looks effective in isolation becomes hard to scale across an enterprise that is not aligned in the same direction.”
– Forbes, Probert and Forooghiano

For resilience engineering teams, the implication is clear: validating AI systems requires enterprise-level coordination across data, governance and operating models.

As banks embed AI into customer journeys, payments systems and risk models, the operational consequences of poorly integrated transformation programmes become increasingly significant.

Testing teams must now validate complex interactions between data platforms, AI models, legacy systems and digital channels. Failures in these environments can propagate quickly across distributed architectures, increasing the importance of resilience testing and chaos engineering techniques.

The Capco authors argued that organisations must treat AI transformation as a structural shift rather than simply a technology upgrade.

“AI represents another transformation opportunity,” Forbes, Probert and Forooghiano concluded.

However, they did stress that the sector must learn from past experience. “To fully realise its potential firms must learn from the patterns that have impeded previous transformation initiatives.”