AI promises a lot. But when it comes to IBM i applications, the path is anything but straightforward. These platforms have been running quietly – and critically – for decades, often under the radar of modernization efforts. But applying AI to IBM i isn’t just about adding automation or analytics. It’s about preparing the system to be understood. That means tackling challenges many overlook from buried logic in decades-old RPG code, to missing documentation, to disappearing system knowledge as key personnel retire. In this article, we’ll walk through three foundational gaps you must close before AI can truly deliver meaningful value on IBM i (AS/400). In upcoming posts, we’ll also explore how companies can address these gaps – without rewrites, disruptions, or risky shortcuts.
1. Decoding Legacy Logic That Outlived Its Authors
Many IBM i applications were built and modified over years – even decades – without proper version control or documentation. The result: applications that work, but few fully understand. For AI to support modernization, the first step is untangling this opaque structure. That means identifying modules, mapping dependencies, and surfacing business logic hidden deep in procedural code. AI can assist in this decoding. But only when designed to handle IBM i’s unique legacy landscape.
2. Rebuilding Business Context from Code Alone
Knowing what the code says is different from knowing what the system means. Business rules, customer-specific logic, and operational exceptions are often hard-coded without clear traceability. Generic AI tools fall short here. What’s needed is an intelligent layer that connects the technical with the operational – bridging code with the business processes it supports. This step transforms analysis into actionable knowledge, making AI outputs truly relevant to business and IT stakeholders alike.
3. Delivering Usable Knowledge, Not Just Outputs
Even with deep analysis, AI adoption depends on how clearly it communicates. Engineers and managers don’t need another black box. They need documentation, interactive views, and concise answers they can trust. For IBM i (AS/400)., this means converting raw insights into user-friendly representations: data flows, object relationships, and plain-language explanations. AI should serve as a guide, not just a generator.
What’s Next
Legacy doesn’t mean obsolete, but applying AI to IBM i means starting with the right foundations. In the next articles, we’ll introduce how companies can reimagine this entire process using a purpose-built AI platform – one that brings clarity, speed, and confidence to the teams managing aging yet mission-critical systems. Stay tuned.
By Kai Wiens
By Ivo Sokolov