Solving the “Messy Middle” of Digital Transformation

Solving the “Messy Middle” of Digital Transformation

Digital transformation is rarely clean, and legacy systems don’t disappear overnight. In fact, they persist because they remain mission-critical to industries like healthcare, finance, and government. At Prodacity, Nick Keune, Head of Pre-Delivery at Mechanical Orchard, explored the reality of modernizing legacy mainframes and where AI fits into the equation.

AI Alone Won’t Solve Legacy Modernization

There’s a misconception that AI will magically untangle the complexities of legacy system modernization. The reality? AI can’t do better than what humans already understand, especially in systems where failure is not an option.

Keune pointed to high-stakes projects like AstraZeneca’s COVID vaccine adverse event reporting and Project Lightspeed’s clinical trial data. These workloads rely on mainframes and legacy systems, and introducing non-deterministic AI models into such environments presents a trust gap. If an AI “hallucinates” a new business rule, the consequences can be catastrophic. Would you trust AI to write new tax regulations? What about to interpret existing ones?

For developers, this highlights a critical distinction: AI isn’t a replacement for human judgment, it’s a tool that requires strict validation before it can be trusted in production systems.

The Economic Reality of Legacy Maintenance

Legacy technology isn’t just a technical debt problem, it’s an economic one. Organizations spend over $1.1 trillion (some data points state as high as $13 trillion) annually maintaining legacy systems, a figure that has significant implications for global GDP.

A 2019 Accenture survey found that nearly half of global banking systems still rely on legacy technologies, with many built on COBOL, a programming language developed over six decades ago. This outdated foundation presents significant challenges, from security vulnerabilities to difficulties in integrating modern digital services.  

The financial burden of maintaining these aging systems is staggering. According to a 2020 Capgemini report, banks in North America and Europe allocate as much as 75% of their IT budgets to sustain legacy infrastructure, leaving little room for technological advancements and innovation. This imbalance slows digital transformation efforts and makes it harder to keep pace with evolving customer demands.

One example Keune shared was a quarter-million-dollar quote for a project to add a single healthcare error message – simply to explain why a claim was denied. This illustrates the unsustainable cost structures of legacy maintenance. The problem isn’t just that old systems persist, it’s that modifying them carries unjustifiable costs and long ROI cycles.

Incremental Modernization with Behavioral Twins

Rather than pursuing high-risk, high-stakes migrations, Keune emphasized incremental modernization explained as a process that mirrors legacy system behavior exactly before making changes. This approach consists of using AI tools alongside three core principles:

  1. Treat the running system as the specification – Instead of relying on outdated documentation, capture how the system behaves in production.
  2. Reverse engineer a behavioral twin – Create a new system that produces identical outputs before introducing changes.
  3. Deploy incrementally – Validate equivalence at each step, allowing cutovers to happen seamlessly.

This strategy removes the guesswork and risk from modernization. Instead of blindly trusting AI to generate code, developers should verify AI outputs against real-world system behavior. The goal? Add nothing, change nothing, and delete everything unnecessary.

 What This Means for Developers and the Industry

The industry is moving away from risky “big bang” migrations in favor of incremental cutovers, a strategy that minimizes disruption and builds confidence in modernization efforts. Rather than attempting large-scale system overhauls, organizations are increasingly adopting behavioral twins, a method that replicates legacy system behavior within modern architectures before making changes. This approach ensures continuity while allowing gradual improvements.

AI alone isn’t the answer to modernization. The real value lies in AI-powered validation frameworks that help developers test and verify modernized systems against their legacy counterparts. Rather than replacing human expertise, these tools should act as safeguards, ensuring that updates and migrations don’t disrupt critical workflows. Without effective validation, AI-generated code risks introducing new inefficiencies instead of solving existing problems.

With organizations spending trillions annually on legacy system maintenance, the financial burden of outdated infrastructure is becoming impossible to ignore. Businesses can no longer afford to allocate the majority of their IT budgets to upkeep while neglecting innovation. The industry must shift toward cost-effective modernization strategies that balance risk, efficiency, and long-term viability rather than continuing to invest in systems that hinder progress.

Final Thought

The messy middle of digital transformation isn’t going away, and AI won’t be a magic bullet. But by embracing incremental modernization, rigorous validation, and behavioral twins, organizations may modernize without taking on unnecessary risk.

For developers, the takeaway is clear: Trust AI, but verify everything – you are accountable.

Authors

  • Paul Nashawaty, Practice Leader and Lead Principal Analyst, specializes in application modernization across build, release and operations. With a wealth of expertise in digital transformation initiatives spanning front-end and back-end systems, he also possesses comprehensive knowledge of the underlying infrastructure ecosystem crucial for supporting modernization endeavors. With over 25 years of experience, Paul has a proven track record in implementing effective go-to-market strategies, including the identification of new market channels, the growth and cultivation of partner ecosystems, and the successful execution of strategic plans resulting in positive business outcomes for his clients.

    View all posts
  • Bringing more than a decade of varying experience crossing multiple sectors such as legal, financial, and tech, Sam Weston is an accomplished professional that excels in ensuring success across various industries. Currently, Sam serves as an Industry Analyst at Efficiently Connected where she collaborates closely in the areas of application modernization, DevOps, storage, and infrastructure. With a keen eye for research, Sam produces valuable insights and custom content to support strategic initiatives and enhance market understanding. Rooted in the fields of tech, law, finance operations and marketing, Sam provides a unique viewpoint to her position, fostering innovation and delivering impactful solutions within the industry. Sam holds a Bachelor of Science degree in Management Information Systems and Business Analytics from Colorado State University and is passionate about leveraging her diverse skill set to drive growth and empower clients to succeed.

    View all posts