The global banking sector is currently navigating a period of profound structural transformation, characterized by an escalating conflict between decades-old legacy infrastructure and the urgent mandate to adopt artificial intelligence. For years, the narrative surrounding financial technology has focused on the agility of neo-banks and fintech startups, which operate without the "technical debt" of their traditional counterparts. However, as 2024 and 2025 have unfolded, the conversation has shifted from simple digitization to a more fundamental requirement: the need for absolute operational transparency. Traditional institutions, many still reliant on mainframe technologies and COBOL-based systems that process the majority of the world’s daily transactions, are finding that they cannot implement advanced AI without first understanding the "brutal reality" of their existing internal processes.
The Reality Gap in Financial Operations
The core challenge facing modern banking is not merely the age of the systems but a pervasive lack of visibility. While executive boards often operate based on idealized process designs and theoretical workflows, the actual data suggests a significant divergence between expectation and reality. According to Werner Horn, founder and CEO of ProcessLab and a Celonis Platinum Partner, when management teams are presented with real-time data regarding their operations, the initial reaction is frequently one of denial. Horn, who has spent over 25 years in process management, notes that the discrepancy between how a process is designed and how it actually functions is often stark, leading to a "brutal" realization for leadership.
This "reality gap" is exacerbated by what Horn describes as "lazy opinions"—a tendency within large organizations to build governance and process improvements around anecdotal evidence or high-profile individual failure events. Instead of looking at the holistic data picture, banks have historically reacted to specific exceptions. This reactive posture results in a patchwork of manual workarounds and "shadow" processes that remain invisible to senior leadership until surfaced by process mining and intelligence tools.
Recent research underscores the scale of this problem. A study published in early 2026 involving senior practitioners across the financial services sector revealed that 94% of organizations still struggle with siloed data, despite years of heavy investment in integration projects. Furthermore, the study found that between 30% and 70% of professional time in these institutions is consumed by manual data assembly and reconciliation. One financial leader described the burden as "doing an MBA homework assignment" every time a report is prepared for the C-suite, highlighting a system where assembly takes precedence over actual analysis.
A Chronology of Forcing Functions: From Savings to Survival
To understand why the banking industry is only now prioritizing process intelligence, it is necessary to look at the historical "forcing functions" that drove other sectors. In manufacturing, supply chain management, and procurement, thin margins necessitated a granular focus on efficiency. For decades, these industries fought for incremental gains of 1% or 2%, as these margins often represented the difference between profit and loss.
Banking, by contrast, operated under different economic structures and competitive dynamics. For much of the early 21st century, the incentive to dismantle complex, opaque processes was not strong enough to outweigh the perceived risk and cost of core system replacement. Chris Johnston, SVP and Head of Global Banking at Celonis, observes that until roughly five years ago, process work in banking was largely "nascent" and "retrospective." It was a diagnostic tool used for point-in-time looks at singular processes, rather than an operational necessity.
The timeline of change accelerated with the arrival of "agentic AI"—AI systems capable of not just analyzing data, but taking autonomous actions. This technological shift has transformed the business case for process intelligence from a quest for incremental savings into an existential mandate. Banks now recognize that to compete with the seamless, automated customer experiences offered by digital-native competitors, they must reimagine their core operations, from loan origination to customer onboarding.
Bridging Technical Debt with Composable Architecture
The traditional solution to legacy technology—the "rip and replace" strategy—is increasingly viewed as a high-risk, prohibitively expensive path that few institutions can afford to take. The disruption caused by migrating core banking platforms often leads to multi-year projects that are obsolete by the time they are completed.
The alternative currently gaining traction is the deployment of process intelligence as a "composable layer." This approach allows banks to connect to their existing systems of record and engagement without dismantling them. By using platforms like Celonis to trigger "light actions" and automations across a legacy stack, traditional banks can mimic the agility of a neo-bank. This architecture "hops" over the depths of technical debt, targeting specific use cases where action is required rather than attempting to refactor the entire underlying infrastructure. This enables institutions to act as though they are free from the constraints of their legacy systems while maintaining the stability of their established foundations.
Case Study: Standard Bank’s Digital Twin
The practical application of this strategy is best illustrated by Standard Bank, the largest financial institution on the African continent. Operating across more than 20 countries, the bank faced significant challenges with cross-border payments—a process that was historically fragmented, manual, and opaque.
In 2022, Standard Bank partnered with ProcessLab and Celonis to create a "digital twin" of its cross-border payment operations. At the start of the project, payment processing times could reach as long as 55 hours, with little visibility into where the bottlenecks occurred. By mapping the data in real-time, the bank identified specific friction points across its international markets. The results were transformative:
- Turnaround Times: Reduced from 55 hours to a handful of hours.
- Straight-Through Processing (STP): The rate of automated, intervention-free processing reached over 90%.
- Operational Rigor: The bank moved from a diagnostic view to a real-time operational management model.
This success has led to the development of standardized products aimed at the wider banking market. New applications, such as the Home Mortgages Manager and the FSI Customer Service & Experience Manager, are designed to provide mid-market and upper-mid-market banks with a faster route to these efficiencies. For example, the mortgage application creates a living digital twin of the loan journey, from application to funding, specifically targeting the manual rework loops and handovers between attorneys and appraisers that typically delay approvals.
The Human Element: Resistance and Change Management
Despite the technological advancements, the primary barrier to transformation remains human. The introduction of process intelligence often creates friction within the "middle management" layer of an organization. When processes become transparent, it becomes clear which areas are underperforming and where manual effort is being wasted.
Werner Horn notes that process intelligence can surface productivity at a resolution that shows output by person, a level of visibility that can be perceived as an existential threat. "The reality is quite brutal when they see what’s actually going on," Horn says. In some instances, simply implementing the technology and making it clear that there is "no place to hide" has resulted in a 20% to 30% uplift in productivity before any actual process redesign was even implemented.
Furthermore, the data often reveals surprising inefficiencies. In one analysis of a bank’s customer service operations, it was discovered that approximately 12% of all incoming query volumes (calls and emails) were merely "thank you" notes. Under legacy systems, staff were required to manually open and act on each of these. By automating the classification and routing of these interactions, the bank was able to significantly reduce response times and improve adherence to Service Level Agreements (SLAs).
Implications for the Global Banking Landscape
The shift toward process-led transformation suggests a new era of competition in the financial sector. As AI success rates remain a challenge—with only 21% of organizations reporting success rates above 80% due to poor data quality and disconnected systems—the ability to ground AI in a "process-first" context will likely be the deciding factor in the coming years.
The path forward for institutions carrying decades of technical debt is no longer about the wholesale replacement of hardware, but about the strategic application of intelligence. By focusing on complex, high-impact areas—such as mortgage lending or cross-border payments—and securing executive sponsorship for end-to-end visibility, banks can create "concrete, demonstrable stories" of success.
As the industry moves toward an agentic future, the "forcing function" is no longer just about saving cents on a transaction; it is about the fundamental ability to deliver a "delightful" customer experience in an increasingly automated world. For the traditional bank, the choice is clear: confront the brutal reality of the data today, or risk being outpaced by those who have already built the digital twins of their operations. The destination of AI in banking may still be evolving, but the roadmap is increasingly being written in the language of process intelligence.
