Rewriting the outcomes

Mainframe systems have long been the backbone of critical operations across industries such as banking, insurance and retail. However, over the decades, these systems have become increasingly complex, costly and challenging to maintain. The original architects and maintainers are retiring, leaving behind systems that are often poorly documented and resistant to change. This presents a significant obstacle for organizations aiming to adapt in a rapidly evolving economic and regulatory landscape.

Recent incidents underscore the urgency of improving the ability of critical systems to adapt. Between 2023 and early 2025, major UK financial institutions experienced over 33 days of IT outages attributed to mainframe failures. Airlines with aging systems have had incidents and issues impacting the global movement of goods and people. These situations not only resulted in substantial financial losses but also eroded customer trust and damaged reputations.

And yet, traditional modernization approaches — such as translating old code into modern code or "lift and shift" strategies — often fall short. They tend to produce systems that are just as cumbersome and difficult to maintain as their predecessors. What’s more, these projects are fraught with risks, including nebulous timelines, escalating costs and uncertain outcomes. Consequently, many organizations hesitate to embark on modernization journeys, opting instead to maintain the status quo despite the risks. And those that do often choose short-term technology goals like “get x application into the cloud,” rather than orienting around business outcomes.

What “good” looks like

Large organizations are immensely complex and always in motion. Technology leaders working to tune systems to the needs of the business constantly evaluate and make tradeoffs between operational needs today and capturing opportunities in the future. In this harried environment, it can be easy to lose sight of what “good” modernization looks like.

Let’s take a step back and broaden our perspective: we propose three characteristics of “good” modernization:

Prioritized based on what matters to the business

Good modernization starts by aligning technical evolution with business priorities. Not every system needs to be modernized, and certainly not in the same way or on the same timeline. Efforts should be prioritized on the systems that are directly impacting competitive advantage or that actively hold the business back. 

This requires close collaboration between technical and business teams to assess where modernization will unlock the most value — whether that's accelerating product delivery, improving customer experience or enabling strategic flexibility. The goal isn’t technical purity — it’s making deliberate choices about where and how to invest so that modernization efforts directly serve the organization’s objectives.

Built for lowering the cost of change in perpetuity

A hallmark of good modernization is the ability to evolve continuously without major disruption. This means designing systems and operational practices in ways that minimize the cost of change with, for example, modular architectures, automation and clear interfaces.

The result is a technology stack on which the organization can respond quickly to changing conditions without introducing new fragility or slowing execution. Modernization doesn’t just improve system resilience — it gives teams the confidence and flexibility to experiment, iterate and adapt even in the most unpredictable and fast-moving environments.

Oriented around opportunity gains and cost reductions

Modernization is often framed in terms of cost savings or risk reduction, but that’s an incomplete view. Good modernization also creates “offensive” benefits — unlocking the ability to enter new markets, spin up new offerings or adapt business models more quickly. 

Capturing the full value of modernization may require a mindset shift: from measuring return on investment purely through backward-looking efficiency metrics to recognizing the forward-looking gains that come from increased agility and speed. Organizations that evolve their ROI models to account for these opportunity-driven outcomes are better positioned to make strategic technology decisions that fuel long-term growth.

A fresh approach to an old problem

In response to these challenges, Thoughtworks and Mechanical Orchard have formed a strategic partnership to rethink the approach to mainframe modernization from first principles. 

A crucial recognition was that common failure points included technology elements like the prohibitive nature of verification, and the socio-technology elements like stakeholder input, documentation and organizational context.

Based on this observation, Mechanical Orchard built a mainframe modernization platform called Imogen that focused on system behavior (represented by code and data), based on tooling developed while modernizing customer mainframes. Using data capture agents, Imogen is able to define how an application works as part of the system: its interactions with other applications, CICS screens, databases and so on. This aligns with and compliments Thoughtworks’ socio-technical philosophy toward modernization and helped form a joint offering around three concepts:

Incrementalism

Replacing a mainframe system all at once is risky and often disruptive. Taking an incremental approach — modernizing in small, manageable steps — helps teams avoid major disruptions and keeps business operations stable. Each successful step builds confidence across the organization and earns trust from stakeholders.

Replicate what’s working first

Many mainframe systems, for all their age and complexities, continue to deliver reliable and mission-critical functionality. Replicating the system’s behavior exactly respects the decades of embedded logic and business rules encoded in these systems, while moving towards modern architectures and creates a stable, verifiable baseline on which to apply generative AI-based tools. The “data capture first” approach represents a unique shift-left model in creating a method of rapidly testing new code against actual system behavior.

Working software, always

Plans and diagrams are helpful, but working software speaks louder. The clearest sign of real progress is code that runs, passes tests and performs reliably in real-world scenarios. This principle, drawn from agile development, applies with even greater force in mainframe modernization, where success is often measured in continuity and correctness. These tangible results help reduce uncertainty and strengthen trust in the overall modernization effort.

Smarter and safer

For decades, modernizing mainframe systems has been viewed as too risky, too expensive, and too uncertain, only to be undertaken when systems are literally falling apart. Moving forward with traditional approaches often produce underwhelming results.

But that’s changing. Thoughtworks and Mechanical Orchard are challenging outdated approaches with a new model that makes “good” modernization achievable at minimal risk to operations, stakeholders — and careers. By focusing on incremental change, replicating proven functionality, and measuring progress through working software, we’re dramatically reducing risk while unlocking long-term agility and growth.

It’s now possible to do the right thing and get better outcomes, no tradeoff required.