01. Aug 2025

From old to new: Agentic AI as the key to IT modernization

Without IT modernization and AI, companies become inflexible due to legacy systems—as shown by the latest Lünendonk study 2025. But what possibilities does Agentic AI offer to facilitate the replacement of outdated software systems? Find out how Agentic AI helps make legacy systems fit for the future.
1060 x 710 Thomas Jäger

Author

Thomas Jäger

2025 Agentic AI Alt zu Neu Schlüssel

The latest Lünendonk study for 2025 sums it up: IT modernization is one of the most important issues for the coming years – for companies and IT service providers alike. The pressure is growing, as many companies are struggling with outdated software systems that are business-critical but technically obsolete and difficult to maintain. In addition, the necessary expertise for these legacy systems is becoming increasingly rare, while security risks and costs are rising. According to the study, 86 percent of IT service providers see a rapidly growing need for IT modernization – and expect AI functionalities to become standard in the future. 

IT modernization is more than just replacing outdated software systems 

It is the key to ensuring innovation, security, and efficiency in the company. In addition to the risks of old technologies, the pressure to innovate is growing: digital business models, rising customer expectations, and the desire for agility require flexible, modern IT landscapes. AI agents can already effectively relieve the burden of these tasks today:

  • Automatic post-documentation: AI creates missing or outdated system documentation.
  • Intelligent chatbots: Quick access to expert knowledge and technical assistance.
  • Code transpilation: Automatic translation of legacy code, e.g., from Cobol to Java.
  • System and dependency analysis: AI recognizes relationships and critical components.
  • Code refactoring: Improvement and simplification of existing code bases.
  • Test case generation: Automatic creation of tests for better code coverage and quality assurance.
  • Automated data migration: Reliable data transfer through mapping and cleansing.

Despite all the automation, human expertise remains indispensable

Software engineers monitor the work of AI agents, make critical decisions, and ensure that solutions are optimally aligned with business objectives. The combination of AI and human know-how is the key to successful transformation.

Our experience from current customer projects shows how this interaction between human expertise and agentic AI works in practice.

Practical examples of IT modernization with AI

Microservices migration from Node.js/JavaScript to Kotlin/Spring Boot in the government environment

In a recent customer project, the project team migrated microservices from Node.js/JavaScript to Kotlin/Spring Boot – supported by AI agents such as GitHub Copilot and Cursor AI. The project goal was to hand over the migration process to AI as much as possible and only intervene where human judgment was required. AI took over code analysis, service generation, test automation, and documentation. 

The project team relies on agentic coding: AI agents analyze the legacy code, extract the business logic, and automatically transfer it to a new microservices structure. In the process, REST APIs, data accesses, and extensive tests were generated to ensure functional equivalence. The new application was containerized and seamlessly integrated into the customer's Kubernetes infrastructure. Automatically generated documentation and migration guides facilitated knowledge transfer.

Analysis and documentation of Oracle PL/SQL in a government agency

In a public sector project, an AI-based prototype was used to maintain and modernize a complex PL/SQL system for government grant applications. The agent analyzed and documented the code, which had grown over many years, automatically and locally – an important step, as maintenance had previously required a great deal of manual expertise.

The use of AI provided a quick overview of relationships and dependencies, reduced the effort required for documentation, and provided targeted support for the further development of the system. LLMs can thus also provide valuable assistance in the maintenance of PL/SQL legacy systems – provided they are used in a targeted manner and in compliance with data protection regulations. 

Migration of Oracle Pro*C code to Kotlin/Spring Boot in an industrial company

For a leading vehicle manufacturer, we are migrating the central production control system from a historically grown Pro*C application to a modern, cloud-native architecture with Kotlin/Spring Boot. The existing solution was technically outdated, difficult to maintain, and not suitable for current container environments.

The result: faster migration, fewer errors, and sustainable modernization. However, our colleagues noted that a convincing result could only be achieved through a combination of clear goals, human review, and targeted use of AI.

The challenges for agentic coding are still manifold

Our experience shows that even with AI, modernization remains challenging. Typical challenges currently being discussed by AI experts include:

  1. How do you create agentic workspaces in an existing code base? Clear interfaces and structures must be established so that AI agents can be effectively integrated.
  2. How do you migrate code efficiently? There are various approaches here: In the code-to-code method, source code is translated directly into a new language or platform. The code-to-doc-to-code approach first documents and abstracts the existing logic before creating modern code from it – particularly valuable for complex or poorly documented systems.
  3. How do you break down large, monolithic systems for modernization? It's about dividing systems in a way that allows individual components to be migrated or modernized in a targeted manner by an AI agent.
  4. How do you deal with rare programming languages that have not yet been trained in language models? Creative strategies are needed here, such as human expertise, specialized tools, or hybrid approaches.

IT experts must analyze code, identify dependencies, and create migration plans – all without jeopardizing ongoing business operations. Agentic Coding supports this by using AI agents to automate data migration, optimize system integration, and minimize sources of error. Using advanced algorithms, they accelerate the entire modernization process and ensure seamless transitions between old and new systems.

IT modernization with AI is the key to future-proofing

The Lünendonk Study 2025 and our project experience show that Agentic AI is not an end in itself, but a powerful tool for IT modernization. The clearer the goals and the better the collaboration between humans and AI, the more successful the transformation from old to new will be.
Are you currently facing the challenge of modernizing your own legacy systems and need an experienced partner to accompany you on this journey? We look forward to discussing this with you. 

Shape your future with Accso AI-Native: For us, this means integrating artificial intelligence into every aspect of the software lifecycle. From initial conception to continuous optimization, we combine state-of-the-art AI technologies with decades of development expertise to help you achieve your business goals faster, more efficiently, and more sustainably. With an experienced team of AI specialists and software architects, we develop customized solutions that are perfectly tailored to your business requirements.

Learn more about AI-Native at Accso or contact us directly.

Select contact

Thomas Jäger

Your contact for all questions regarding software engineering and operations at Accso.
Thomas Jäger Raute