Legacy systems, such as those, built with COBOL and PL/SQL, or outdated Java frameworks and other ageing technologies are still part of the IT landscape and still drive vital operations in several industries. Yet, these applications have gradually become technologically obsolete, and this is limiting business innovation. Rising running costs, limited scalability, security issues, and a shrinking talent pool make transformation urgent. Manually rewriting the code for application transformation, however, can be time-consuming and expensive, while also introducing a high risk of human error.
In this context, GenAI offers a game-changing opportunity to accelerate automation to modernise these systems efficiently and risk-free. There is one important caveat, however, and that is to unlock GenAI benefits and achieve effective application transformation, organisations require specialised expertise, purpose-built tools, and well-defined, methodical processes.
The role of GenAI in application modernisation
With the rise of GenAI, application modernisation can in fact be approached semi-automatically, reducing costs, time, effort, and risks. By iteratively transforming code, designing new architectures, and generating documentation and test cases, GenAI enables faster, more accurate transformations. This approach not only ensures consistency across the modernised system but also allows the measuring of key performance indicators (KPIs) throughout the project, making legacy system migrations more trackable, and manageable, with staggering potential cost savings of 30–60% compared to traditional methods.
The four key steps to apply this approach successfully are:
- Analysis anddiscovery
The success of a GenAI-driven application modernisation project also relies on a precise plan and a well-structured approach, which starts with a phase of analysis and discovery. This represents a foundational cornerstone which involves an in-depth assessment of the legacy environment, including the source application structure and the classification of its components based on their role within the system.
This phase as critical since it requires not only scanning, but also categorizing, cleansing, and mapping source code to define transformation units. Moreover, this step also entails establishing a clear logical mapping between source and target components.
During this phase, the target architecture is also designed by selecting frameworks, patterns, and best practices, such as choosing between monolithic or modular structures.
The analysis is further strengthened by evaluating code volumes, typologies, and component composition, helping identify unused portions of code that should be excluded from the modernisation project scope. By profiling the existing architecture and understanding how the system is used, organisations can more effectively prioritise what to update, or re-architect.
- Prompt Engineering and Migration Engine Configuration
The second phase focuses on the design and development of the prompts that will drive the transformation of the source code into the target programming language, architecture, framework, and guidelines. This stage establishes the foundation for semi-automated modernisation, ensuring that transformation follows the predefined design and adheres to the target system’s standards. It includes the configuration of the prompts’ pipelines, which involves grouping multiple prompts and orchestrating their execution for efficient and accurate code transformation.
By carefully engineering and structuring these prompts, organisations can guide GenAI to produce consistent, high-quality outputs. This structured approach reduces the risk of errors, accelerates the migration process, and ensures that the transformation is both scalable and repeatable across large legacy systems.
- Iterative Code Modernisation Runs
This is a critical phase consisting of multiple iterative runs of the modernisation process, each designed to progressively transform the legacy code. It is divided into three steps, beginning with the execution of software artifacts to pre-process and analyse the code, and then adding comments to identify specific patterns that will guide the transformation.
Automatic migration runs follow and generate the target application, transform the source code according to the engineered prompts, produce technical documentation, and create unit test cases. Finally, post-processing tasks analyse the transformed code to identify opportunities for optimisation and refinement, while KPIs are measured to monitor the quality of the transformation.
- Testing and Consolidation
Once the best possible result is delivered, this final phase, carried out by software development team, ensures the target application is fully verified and ready for production. It typically includes comprehensive technical testing, functional testing, and user acceptance testing (UAT), along with the implementation of any necessary corrective actions.
Stress and performance tests, as well as penetration tests, are also planned and executed to ensure the robustness, security, and reliability of the modernised system. Depending on the rollout strategy, the code can then be promoted to the production environment, sometimes following a parallel deployment phase to minimise operational risks.
Conclusion
The success of GenAI-driven application modernisation is made possible by proprietary engines that orchestrate and accelerate the transformation process. This approach significantly reduces costs and project timelines, and streamlines developers’ refining, integrating, and testing work.
With structured methodologies, know-how and tools and the support of experienced system integrators organisations can modernise legacy systems confidently, eliminate outdated technologies, and achieve substantial cost savings while enabling a cloud-ready, scalable, and reliable architecture. This provides CIOs with a concrete path to lead their teams and users towards future-ready, innovative IT applications.
