Vogel Consulting LLC

IBM Bob and the Old Programmer Who Finally Climbed Aboard

I have been hearing that the AS/400 is obsolete since Y2K.

Back then, everyone was talking about abandoning it for the next big thing — new platforms, new frameworks, new acronyms. Yet somehow, the so-called “obsolete” machine kept quietly running businesses all over the world. I did not jump ship. I kept sailing.

Today it is called IBM Power Systems running IBM i. The name changed, but the core idea did not: reliability, integration, and business computing that simply works. That “i” officially stands for integrated. I have another word for it: innovative.

An Honest Skeptic

Call me a critic. That is fair.

I have had my share of AI snafus — nonsense pie charts, polished but completely fabricated explanations, confident babble with no connection to accurate information. So yes, I approached AI cautiously.

Listen to the advertisements today and it sounds like “I, II, III, AI, AII, AIII, IAIAI, AIAI” on a loop. It reminds me of that old joke where people dramatically chanted Roman numerals:

“Eye. Eye-eye. Eye-eye-eye. Eye-vee. Vee. Vee-eye…”

I have had enough hype and acronym overload. I want to see real results.

The Deep Dive

I have not been SCUBA diving since September 2025, but that does not mean I have not done a deep dive.

I spent months testing AI against real-world development work — not flashy marketing demos. That process revealed plenty of frustrations. But recently, something changed.

I am now officially all aboard with my friend IBM Bob.

(I am also getting acquainted with IBM TOBi, but that is a story for another day.)

33 Years, Three Rewrites

In 1993, I wrote a standard maintenance application modeled after the style and behavior of IBM Programming Development Manager, built with subfiles. Like most RPG developers of that era, the design reflected the standards and techniques of the time.

In 1999, I rewrote it.

In 2026, I rewrote it again — this time in modern free-format RPG.

But this rewrite was different. Instead of simply modernizing syntax, I used AI to help analyze decades of accumulated design decisions, technical debt, and modernization opportunities. I wanted to see if AI could understand real IBM i development problems, not just produce generic textbook advice.

So I put IBM Bob to work.

Putting IBM Bob to the Test

I gave it the same practical challenges experienced RPG developers face every day:

  • Legacy architecture
  • Subfiles
  • Global variables
  • Subroutines versus sub-procedures
  • Encapsulation and modularity
  • Maintainability
  • Modern free-format RPG standards

These were not toy examples. This was real code with real history behind it.

My first prompt laid out the full context — the original 1993 design, the 1999 revision, the recent free-format rewrite, and the architectural concerns I wanted addressed, particularly the overreliance on global variables and the need for better procedure design.

My second prompt was simple: “The file is located in a different directory…”

That may not sound exciting, but anyone who has worked with real development projects knows that context matters. AI systems often fall apart when dealing with real-world file structures. IBM Bob followed along without missing a beat.

Then I asked the question that matters most with any AI tool:

“Did you rewrite the code with your suggestions, or did you just make suggestions?”

That is where many AI tools fail developers. They become spectators instead of contributors.

IBM Bob did not stop at commentary. My next instruction was direct: “Switch to Code mode and start implementing the refactorings.”

And it did.

Not perfectly. Not magically. Not without review. But productively.

What Actually Impressed Me

It was not that IBM Bob generated code. Plenty of systems can do that.

What impressed me was that it understood the intent behind modernization efforts in an IBM i environment. It recognized architectural patterns. It identified maintainability issues. It grasped why moving from global variables toward parameter-driven procedures matters — not just syntactically, but practically.

More importantly, it translated decades of RPG experience into something that modern developers working in tools like Visual Studio Code can actually follow and build on.

That matters. The future of IBM i development does not depend on pretending the platform is frozen in 1995. It depends on making decades of business logic accessible, maintainable, and understandable for the next generation of developers.

That is where I finally started seeing real value in AI.

Not replacing developers. Not replacing experience. Not replacing critical thinking. But accelerating modernization and helping bridge generations of technology and programming practice.

The Irony

After 25-plus years of being told the AS/400 was obsolete, one of the most practical uses of AI I have experienced has been helping modernize RPG applications running on IBM i.

Maybe the old ship was worth staying aboard after all.

View All

Vogel Consulting LLC