For many organisations, IBM Power Systems, running IBM i (formerly AS/400) or AIX, represent decades of mission-critical investment. The applications they support have processed millions of transactions, underpinned entire business models, and quietly kept the lights on for industries ranging from manufacturing and distribution to financial services and the public sector.
But the world around those platforms has changed dramatically, and the pressure to modernise is now acute. Agile development cycles, cloud integration, API-first architectures, and modern user experience expectations are hard to reconcile with applications written in RPG or COBOL across multiple decades of iterative development, often with little documentation and fewer people who understand them deeply.
The question for most IBM Power customers is no longer whether to modernise, but how to do it without disrupting the operations that depend on those systems every day. Increasingly, AI is changing the answer to that question.
The question is no longer whether to modernise IBM Power applications, it's how to do it without disrupting the operations that depend on them every day. Share on XThe legacy problem: more complex than it looks
IBM i and AIX applications are often described as legacy, but that word understates the challenge. These aren’t simply old systems; they are deeply embedded systems. Business logic has accumulated over thirty, sometimes forty years, with rules, exceptions, and workarounds baked into the code that no single person fully understands anymore. In many cases, the original developers have retired or moved on. Documentation, where it exists at all, is incomplete, inconsistent, or simply wrong.
The result is what practitioners sometimes call a ‘black box’ problem. The application works — and critically, the business depends on it working — but the organisation cannot safely change it, extend it, or connect it to modern systems without significant risk. Every modification becomes an exercise in archaeology, and every release carries the anxiety of unintended consequences.
The technical debt compounds over time. As regulatory requirements evolve, as customer expectations shift toward digital-first experiences, and as integration with cloud platforms becomes a commercial necessity, the cost of maintaining that status quo rises. Modernisation ceases to be optional.
The skills crisis is accelerating the urgency
IBM i and AIX modernisation has always been a specialist discipline. Today, that specialism is becoming critically scarce. The developer community with deep RPG, CL, and COBOL expertise is ageing, and relatively few younger developers are entering those disciplines. The pipeline of skilled resource is narrowing at precisely the moment demand is growing.
This creates a compounding problem for organisations that delay modernisation decisions. The longer they wait, the fewer people will be available to help them when they do act, and the higher the cost of that help will be. In competitive talent markets, experienced IBM i developers command significant premiums, and even at those premiums, capacity is often insufficient to support large-scale modernisation programmes.
At Northdoor, we work with organisations across sectors who are confronting this reality. The skills shortage is not a future risk — it is a present constraint that is already limiting what modernisation programmes can realistically achieve within acceptable timeframes and budgets.
The developer community with deep RPG, CL, and COBOL expertise is ageing. The pipeline of skilled resource is narrowing at precisely the moment demand is growing. Share on XWhere AI changes the equation
The emergence of AI-assisted development tools, and specifically, their application to legacy code analysis, refactoring, and documentation, represents a genuine step change in what modernisation programmes can achieve.
The core capability is this: AI can read, interpret, and reason about legacy code at a scale and speed that no team of human developers could match. What previously required months of manual analysis, mapping business logic, identifying dependencies, understanding data flows, can now be accelerated significantly, enabling modernisation programmes to move faster, with greater confidence, and with a materially reduced dependency on scarce specialist skills.
AI-Assisted Code Analysis and Refactoring
One of the most immediately valuable applications is automated code analysis. AI tools trained on IBM i and AIX environments can ingest RPG, CL, COBOL and associated programmes, map the relationships between them, and produce structured representations of what the application actually does — not what the documentation says it does.
This matters enormously in practice. When modernisation teams begin working on a legacy estate, the first challenge is always understanding scope and complexity. Which modules are interdependent? Where does core business logic reside versus procedural scaffolding? Which code paths are still active, and which are effectively dead? Previously, answering these questions required extensive manual effort from people who understood the codebase intimately, a resource that is increasingly unavailable.
AI-assisted analysis compresses that discovery phase substantially. It also reduces the risk of modernisation decisions being made on incomplete understanding, one of the primary causes of failed or over-budget programmes.
On the refactoring side, AI tools can assist in translating legacy code into modern equivalents, converting RPG programmes to Java or Python, restructuring monolithic applications into modular, API-accessible components, and identifying opportunities to rationalise duplicated logic. Human oversight remains essential; the AI accelerates and augments the work, it does not replace expert judgement. But the productivity gain is significant, and the reduction in dependency on scarce RPG expertise is commercially meaningful.
AI-generated documentation: closing the knowledge gap
Perhaps the most underappreciated capability is automated documentation generation. For organisations with large IBM i estates, the absence of reliable documentation is often the single biggest barrier to confident modernisation, more limiting, in practice, than the technical complexity of the code itself.
AI tools can analyse legacy programmes and generate structured documentation: functional descriptions of what each programme does, data dictionaries, process flow diagrams, dependency maps, and plain-English explanations of embedded business rules. This documentation serves multiple purposes simultaneously.
It enables modernisation teams, including those without deep legacy expertise, to understand and work with the application. It provides a foundation for testing strategies, ensuring that modernised components can be validated against known expected behaviours. And it creates an institutional knowledge base that the organisation retains, reducing the critical dependency on individual expertise that has left so many IBM Power customers vulnerable.
For organisations that have been operating in a state of knowledge fragility, where the departure of one or two key individuals would represent a genuine operational risk, this capability alone justifies serious attention.