(Subtitle: From Programmable Calculators in the 70s to Co-creating with AI in 2025, the Core Principles of a "Tool Guy" Never Change.)
I recently told an AI assistant that I considered myself a "tool guy," and it sparked a memory from nearly forty years ago—a moment that I now realize has defined my entire career.
It was 1984. I was working on a massive Air Traffic Control system for Riyadh. The project was complex, a distributed system of subsystems all communicating through messages. At its heart was a component called the Main Application Interface (MAI). As the project grew, regenerating this MAI after even a tiny change became a monumental task, taking a full two days to complete. Two days of waiting. Two days of lost productivity.
During one of those long waiting periods, I had an idea. The regeneration process, written in the system's native DCL, was thorough but brute-force. It rebuilt everything from scratch, every single time. It felt inefficient. So, I decided to try to recreate the core logic in a language I knew was faster and more precise: Pure RTL/2.
The result was transformative. The new tool I built regenerated the MAI in minutes, not days.
That was the moment I understood my passion. It wasn't just about writing code; it was about building better tools. It was about finding the inefficiencies in a process and engineering a more elegant, more efficient solution. It was about giving developers back their most valuable resource: time.
This philosophy has been my guiding star for decades. I saw it again when the world moved from slow, batch-oriented compilers to the instant feedback of tools like Turbo Pascal—a revolution in productivity. I see it today in modern IDEs like PyCharm, which feel like a natural extension of a developer's mind.
And now, I'm seeing the next great leap in that evolution: collaborating with Artificial Intelligence.
For the past few weeks, I've been working on a personal project called pi-server-vm. The goal was simple: create a way to test my other big project, PiSelfhosting, without needing physical Raspberry Pi hardware. I wanted a way to spin up a clean, secure, and perfectly configured virtual server in seconds.
The journey to build this tool has been a fascinating dialogue. I provide the architectural vision and the real-world experience, and my AI assistant acts as a force multiplier. It generates the boilerplate code, recalls the obscure syntax for a systemd service file, and helps me brainstorm solutions, allowing me to focus entirely on the system's design. It has become the ultimate tool for the "tool guy."
It feels like that moment back in 1984 all over again. I'm not just writing a script; I'm engineering a system. A system that takes a complex, multi-step process and makes it simple, fast, and reliable.
This blog series will be the story of that project. It's a technical journey, but it's also a reflection on a 55-year career in automation. We'll dive into the code, the bugs we crushed, the architectural decisions we made, and how a developer with decades of experience can partner with the newest technology on the planet to build something they are truly proud of.
In the next post, we'll introduce the pi-server-vm project itself and the problem it's designed to solve. I hope you'll follow along.