AI in IT modernisation has long been more than just hype: it is the driving force behind digital transformation – from analysis through architecture to implementation. Hardly any other technology currently holds as much promise as generative artificial intelligence. In recent years, AI has evolved from an experimental technology into a key component of digital transformation. It can do many things: write code, analyse legacy systems and generate tests. And yet IT managers remain sceptical. This is because it is not only the AI technology itself that poses a challenge, but also its successful integration into the existing IT landscape.
I’m Uwe Barthel, Senior Software Architect at iteratec, and I’ve been working on IT modernisation for many years. In projects, I repeatedly see how AI can help and where it currently still has its limitations. In this article, I would therefore like to show how we at iteratec ensure the use of AI in modernisation projects is realistic, effective and sustainable.
Reality shows that realising the potential of AI in modernisation projects presents companies with considerable challenges. Whilst the technical expertise required to develop and implement in-house AI solutions is already in place, there is often a lack of tried-and-tested methods, a clear implementation strategy and a realistic assessment: where is the use of AI in IT modernisation ? Where does it make sense, and where do the risks still outweigh the benefits?
In my regular discussions with clients, I find that whilst the willingness to change is there, there is often a lack of clear guidance on how AI can actually help within their own IT landscape.
In this article, I would like to show how AI is already being used effectively in IT modernisation projects today: from the analysis of existing systems through architecture development to implementation and migration. In doing so , I will provide insights into our ‘Asset-based Modernisation’process model , which we are specifically expanding and supplementing with AI-supported methods. In this way, we use AI not as a panacea, but as a tool that creates genuine added value.
At iteratec, too, we are still on a journey: we are continuously developing our understanding and our methods relating to AI in IT modernisation. Together with our clients and based on real-world projects.
In our‘ ’ asset-based modernisationmethodology , we combine tried-and-tested methods with innovative technology. The systematic analysis of the current state is carried out, amongst other things, as part of our‘ ’ system health checks and system reviews, with a focus on business strategy, architecture, code quality, performance, security and team structures. This is complemented by the targeted use of AI throughout the four phases of our modernisation approach.
The result: an objective, data-driven overview of the current IT landscape – transparent, strategically relevant and a reliable starting point for further planning.
Figure: The four phases of asset-based modernisation using AI as a tool.
I find the use of AI particularly exciting during the analysis phase, especially in interviews and workshops. We use generative AI to automatically transcribe conversations with stakeholders, cluster content and derive hypotheses from it. We are currently even trialling the active integration of AI into discussions: either as a virtual interviewer to complement traditional interview guides, or as an additional stakeholder who provides data-driven insights and brings new perspectives to the discussion.
The analysis phase is a crucial step in any IT modernisation project. It determines whether a transformation is built on a sound foundation. A deep understanding of the current situation – from a technical, organisational and business perspective – is essential.
However, reliable documentation is often lacking, particularly in established IT landscapes. Knowledge is scattered across the years, has become outdated or has been lost with the departure of experts. This is precisely where generative AI can provide targeted support:
Despite these possibilities, human expertise remains irreplaceable, particularly when evaluating architectural decisions, processes or technical interfaces. Interviews with stakeholders and strategic workshops are indispensable during the analysis to understand the context and relevance. AI complements and supports this process.
“Personally, I find the use of AI in interviews and workshops particularly exciting.”
During the analysis phase, we combine tried-and-tested methods with targeted, AI-supported approaches. A key element of this is our System Health Check, which we use to systematically assess the existing system landscape. In doing so, we examine the following areas: business strategy, architecture, code quality, performance, security and team structure.
Alongside traditional methods, we also utilise AI, for example in the form of an internal analysis tool that automatically identifies dependencies in the codebase and highlights technical legacy issues. We also use generative AI to automatically transcribe discussions with stakeholders , structure the content and develop hypotheses based on this. In pilot projects, we even use AI as an active discussion partner, for example to supplement traditional interview guides or to broaden perspectives through data-driven insights.
The combination of human expertise and AI-supported analysis provides us with a precise, data-driven picture of the current situation – a key building block for any successful modernisation.
Once the current state has been analysed, the conceptual phase of IT modernisation begins with the development of a target vision. In this phase, the direction of the journey – in technological, functional and organisational terms – is defined. AI can support this phase if it is integrated effectively.
Many modernisation projects involve aligning existing systems with future challenges and a future architecture. Which domains will remain in place, and which processes need to be rethought? In this context, we make targeted use of AI agents: AI can assist with interpreting existing documentation, analysing corporate strategy and market trends, and developing new target architectures.
What is particularly exciting is that AI agents canbe deployed specifically to develop decision-making documents, for example to assess target systems, technical feasibility or strategic direction. In practice, teams are increasingly using these intelligent tools as ‘co-pilots’ for architecture workshops or to provide feedback on modelling artefacts.
Personally, I find it particularly fascinating how AI helps, at this stage, to align corporate strategy with current technology trends. It can research relevant market information, generate suggestions and provide data-driven input to the discussion. This not only broadens the perspective, but also often leads to clearer and more transparent architectural decisions.
As part of our strategic design process, we develop the target architecture in collaboration with our clients. Key steps include the strategic assessment of the context, the prioritisation of actions and the formulation of concrete action plans (epics).
In doing so, we combine established methods such as Domain-Driven Design with insights from interviews and workshops, as well as, increasingly, with AI-supported feedback.
Our experience shows that the best visions emerge when technology, business and organisation are shaped collaboratively. AI is not a substitute here, but rather an additional expert who tirelessly analyses, compares and provides new perspectives.
The target architecture is in place; now the implementation of specific technical measures begins. In this phase, technologies are selected, the migration is prepared and the foundations for implementation are laid. It is precisely here that AI is increasingly demonstrating its potential: as an accelerator for repetitive tasks and as a helpful tool in development.
Even though the temptation is great, full implementation via AI is (still) a myth. What does work very well today, however, are targeted applications in clearly defined areas, such as:
It is important to note that AI is not an autonomous development tool, but rather a tool that is particularly helpful when used systematically and supported by experienced teams.
I do not see AI as an autonomous development tool, but as a productive partner, provided it is integrated correctly. In our projects, we rely on an iterative approach with clear architectural models, targeted technology decisions and realistic planning. AI helps us to carry out repetitive tasks more efficiently, so that we as a team have more time for what really matters: sound architectural decisions, creative solutions and technically relevant challenges.
I find the use of AI to validate designs particularly helpful. Rapid prototyping allows ideas to be tested, discussed and adapted within the team at an early stage. This offers real added value when it comes to arriving at viable solutions quickly.
The final phase of modernisation is also the most critical. This phase involves the transition from the old system to the new one without jeopardising day-to-day operations. This is the moment that determines whether the project will work not just on paper, but also in practice. AI can provide targeted support during this phase to minimise risks and speed up processes, provided it is used correctly.
The most common stumbling block in migration is data transfer. Legacy systems often contain data structures that have evolved over time, incomplete data records or outdated formats. This is precisely where AI can provide valuable support:
AI is also used when synchronising processes or interfaces. For example, it helps to identify patterns or inconsistencies.
At the same time, AI can assist in certain scenarios with the refactoring of production-ready components, for example during parallel development and the comparison of behaviour between legacy and new systems.
In many projects, I have found that the pressure during this phase is extremely high. This makes it all the more important to adopt an approach that prioritises stability and clarity. We therefore plan migrations iteratively, in clearly defined steps, accompanied by extensive testing and fallback strategies.
AI does not replace strategic planning. However, it helps me to identify potential sources of error at an early stage, for example through predictive analytics based on historical system data or through simulations in the test system. This provides reassurance for me and all those involved.
Security issues, governance and regulatory requirements play a central role, particularly in large-scale IT modernisation projects. When AI is used as a tool at every stage of IT modernisation, clear rules governing its use are essential: Who makes the decisions? What data is used? How can we ensure that the results are transparent and trustworthy?
That is why we rely on transparent processes, human oversight and a conscientious approach to handling sensitive data. The larger the system landscape, the more important trust, traceability and security become.
However great the enthusiasm for generative AI may be, it too has its limitations. When using AI, one should be aware of these limitations and apply it responsibly in projects. Not all tasks can be automated by AI, and not every problem can be solved with a prompt.
Particularly when working with legacy systems, many AI tools quickly reach their limits: old programming languages, a lack of documentation and highly customised configurations make it difficult for today’s models to fully understand these systems. The idea of modernising an entire legacy system with a single click remains (at least for now) an illusion.
The past few months have shown just how rapidly generative AI technologies are developing and what potential they hold. But what about security? We’ll be exploring the specific considerations involved in a separate blog article soon.
AI can provide valuable support at almost every stage of IT modernisation: analysing, documenting, designing and testing more quickly. And yet: it’s not just about the tool.
What really matters? Experience. Methodological expertise. And a deep understanding of the individual challenges that every modernisation project brings.
I am convinced that AI is a valuable part of our toolbox – especially when combined with realistic expectations, sound methodology and strategic foresight. This is precisely the approach we take at iteratec: not driven by hype, but purposeful, effective and in close collaboration with our clients.
Do you have any questions or need assistance?
Further information on the possibilities for IT modernisation for your business can be found on our website. Feel free to get in touch with us.