Hero imageMobile Hero image
  • Facebook
  • LinkedIn

April 16, 2025

Is Gen AI the way forward for app modernization? Part 3 – Modernizing legacy systems to leverage shifts in technology or reduce technical debt.

Modernization projects are evolving legacy systems to meet the demands of today’s technologies while staying nimble enough to innovate for the future. Developers face a complex reality of technical debt, code inefficiencies, and the growing expectation to deliver faster without compromising quality. For many organizations, the process feels like navigating a maze of mounting deadlines and escalating costs.

When we speak with our clients its very clear that Generative AI (Gen AI) has quickly become a game changer that is redefining how we approach modernization. It’s transforming the developer’s entire workflow and freeing them from cumbersome tasks to focus on creativity. It’s amplifying their craftsmanship skills. 

For organizations, Gen AI offers a smarter, faster path to modernization, and enables teams to scale efforts with precision and collaboration. Future-ready app portfolio enterprises are using Gen AI to empower developers to write better code faster—which accelerates application modernization efforts.

The stage is set for a new era of smarter development.

Gen AI supercharges innovation-ready organizations

Gen AI success is realized when organizations approach its adoption with a strategic lens and integrate it into a broader maturity journey: 

  • Standardization: Documenting and defining best practices.
  • Industrialization: Implementing tooling to improve consistency.
  • Automation: Using advanced tools to handle repetitive tasks.
  • Gen AI: Elevating capabilities through context-aware automation that enables code generation, transformation, and remediation at scale.

Reaching this level of maturity unlocks the true potential of Gen AI code generation, which enables app modernization to become a competitive advantage, rather than a bottleneck. Gen AI helps automate time-intensive tasks, bridge skill gaps, and simplify legacy system complexities, which all lead to the empowerment of development teams to innovate and scale with greater efficiency. The result? Consistent, high-quality outcomes that keep pace with the demands of an ever-evolving digital landscape.

Today’s Coding Challenges

We code to create, to fix, and to evolve. Coding is how we keep applications relevant and reliable. It’s the means through which we innovate with new features and address issues in the tools and platforms we use.

  1. Developing a new featureto meet business needs or create competitive advantages.
  2. Remediating issues—bugs, vulnerabilities, or performance bottlenecks that disrupt operations.
  3. Modernizing legacy systems to leverage shifts in technology or reduce technical debt.

For part 3 and our final scenario we’ll explore we’ll explore why modernizing legacy code due to tech shifts and technical debt is urgent. Cloud migration using PaaS faces compatibility challenges and security risks. And how Gen AI simplifies updates, making modernization strategic and efficient.

Scenario 3: Modernizing the existing code due to tech shifts and technical debt

Consider an example of cloud migration using PaaS. Legacy apps moving to the cloud using PaaS services, like serverless web apps or managed databases, often face compatibility challenges, like old dependencies embedded in the code requiring updates or replacements to ensure smooth operation in the new environment. Apps also need to frequently upgrade outdated software packages to mitigate security risks and prevent breaches.

There is a theme of critical urgency here: technical debt and evolving security landscapes demand immediate action. 

Modernization initiatives set the conditions for developers and IT leaders to grapple with questions like:

  • How do I identify the full impact of these changes before starting?
  • Is the ROI worth the time, effort, and disruption modernization requires?
  • How do I ensure consistency across large-scale upgrades?

For many, the lack of tooling and a clear plan has made these efforts prohibitive in the past. Without a strong rationale for investing time and energy, modernization efforts often stall, or proceed inconsistently.

Why Gen AI is a game changer

Gen AI makes modernization manageable, and, further, strategic. Tools augmented by Gen AI can assess the scope of changes required, and identify dependencies, outdated components, and areas of risk.

They simplify the remediation process by automating fixes directly in the IDE. Platforms like GitHub Copilot and Amazon Q Developer enable developers to make changes efficiently while reducing manual effort.

With the help of Gen AI, modernization goes beyond the scope of updating code — we are talking about planning and executing upgrades with precision, including developing a clear upgrade plan, implementing the remediation process, and testing the changes.

Apps previously deemed unsuitable for modernization due to resource constraints can now be updated with less effort, offering new opportunities for businesses to unlock value.

Our recommendations

Modernizing legacy systems demands a balance of effort, strategy, and innovation. When we talk of transforming outdated applications into secure and efficient systems, we need to bring in a clear approach and the right tools.

Reverse engineer modern assets from legacy code.

For legacy code like COBOL/ASP.NET, Gen AI tools can use reverse engineering to generate the documentation, create user stories, and then convert them into assets matching the target modern language and framework. For Python, app code can be written in Flask or Streamlit frameworks.

As modern frameworks and patterns are significantly different from legacy code frameworks, the approach needs to go beyond one-to-one mapping. We need to take into consideration specific attributes of not only the target coding language but also the web framework and patterns. 

The chat functionality of Gen AI tools can also be used to suggest design orientations like frameworks and design patterns.

When tasks are repeatable, bring in SMEs.

Repeatable tasks, such as software upgrades, dependency management, or specific code transformations, benefit greatly from expert input early in the process. Subject Matter Experts (SMEs) can:

  • Define standardized transformation frameworks that minimize variability.
  • Establish clear guidelines and processes for tasks likely to recur across projects or teams.
  • Evaluate areas where consistency delivers long-term benefits, such as reduced maintenance costs or improved scalability.

SMEs also play a critical role in shaping AI agent capabilities by providing the foundational knowledge that makes automation effective. For example, a language or framework upgrade agent trained with SME-defined standards ensures that some key transformations adhere to organizational best practices.

Scale from developer-led to AI agent-driven.

While developer-driven initiatives can yield quick wins, scaling them across an organization requires a shift to AI agent-driven approaches. 

Identify tasks that are repetitive and prone to inconsistency, such as upgrading multiple websites or modernizing app dependencies. Develop AI agents with clear, reusable prompts tailored to specific tasks, ensuring consistent execution.

When those agents are used by more team members across the organization, these ‘code upgrader agents’ become even more powerful. Imagine the value of a multi-agent system: a ‘code planner agent’, used by all your developers with ‘code upgrader agents’, followed by ‘code reviewer agents’ with multiple autonomous interactions across touchpoints for the best solution.

SMEs would retain ownership of final review and validation of these outputs. It’s the best of both worlds.

Combining our SMEs expertise with GitHub Copilot upgrade assistant for Java or .NET provide a faster, more reliable, and cost-effective migration path by automating code upgrades, reducing manual effort, and leveraging expert insights to ensure optimal code quality and performance.

Figure 1 – GitHub Copilot upgrade assistant for Java

In short: share AI agents and learnings across teams to build an internal ecosystem of tools and practices. Transform isolated efforts into a cohesive modernization strategy.

Preserve developer oversight

While AI agents automate many aspects of code generation and remediation, the final steps—review, refinement, and validation—should remain in the hands of developers. This oversight ensures quality control and accountability, as well as a cycle of continuous improvement with their feedback. 

A programmatic approach for scalability

One of the main pitfalls in modernization efforts is inconsistency. If 10 websites are upgraded using different methods by multiple developers, the lack of standardization creates challenges for future maintenance and scalability.

Gen AI can support consistency with tech-purpose-driven AI agents. Agents can be designed to specialize in specific tasks—such as language or framework upgrades—with mature, reusable prompts that ensure uniform execution across the organization. If organizations share such agents and their capabilities internally, they can capitalize on their learnings and extend consistent processes enterprise-wide.

The broader impact of Gen AI comes from its ability to scale these types of processes across teams and organizations. A programmatic approach ensures that teams adopt consistent practices, using Gen AI tools as part of their standard workflows. Developers can be supported through structured sprints, team coaching, and feedback loops to maximize the value of Gen AI tools. And progress is measured and refined over time, creating a culture of continuous improvement.

Integration matters. Through the IDE, GitHub Copilot can integrate an Atlassian ‘agent’ to leverage ‘in-app’ context of information and knowledge from Atlassian Jira and Confluence.

With an intentional, structured integration of tools, workflows, and practices across teams or the organization, Gen AI becomes a standard part of the development lifecycle. This way, feature development or remediation go from being ad hoc processes to scalable, repeatable frameworks that deliver value consistently.

Embrace the Future with AI-powered Code Generation

Code generation tools powered by Gen AI are truly a game changer for modern software development. They streamline coding processes, boost developer productivity, and unlock new possibilities for app modernization.

Gen AI is, of course, not without its limitations. For example, code transformation from languages like COBOL to Java isn’t yet fully supported by existing tools. In such cases, organizations may need to develop bespoke code transformation solutions to achieve their goals.

Adopting Gen AI tools is only the first step. To fully realize their potential, organizations must elevate their level of discipline—integrating Gen AI into workflows in a thoughtful, structured way. This requires more than just technical capability; it demands a methodology that blends technology, process, and people. Organizations modernizing their apps need to be creating AI agents tailored to specific technical purposes, to deliver repeatable and consistent outcomes. And they need to assemble a comprehensive toolchain to address all aspects of coding, including those activities outside the codebase.

Ultimately, success lies in your ability to evaluate and prioritize scenarios where these tools can deliver the most value. I hope that this series of perspectives serves as a starting point to explore these scenarios, and help you identify where Gen AI can transform your development processes.

According to GitHub Capgemini and Sogeti are the only GSIs they have partnered with for .NET and Java upgrade assistants. By combining our SMEs’ expertise with GitHub Copilot, we offer a faster, more reliable, and cost-effective migration path, automating code upgrades, reducing manual effort, and ensuring optimal code quality and performance.

The future of coding is about working smarter, with scalable processes and sustainable innovation at its core. Now is the time to take that step forward.

Missed the earlier parts of our 3-Part Series?

Discover how Gen AI is reshaping development—automating tasks, bridging skill gaps, and driving innovation to keep businesses ahead. Catch up on Part 1, here.

In Part 2, explore how Generative AI is revolutionizing bug fixes, security enhancements, and performance optimization, making remediation smarter and faster than ever. Read more here.

Pierre-Olivier Patin

Pierre-Olivier Patin

VP Global CTO Applications & Cloud Technologies

Gen AI is a game-changer, transforming how we code, innovate, collaborate and accelerate value delivery.

Pierre-Olivier PATIN
VP / CTO Applications & Cloud Technologies