I spent years on third-level support for operationally critical systems. Flight planning infrastructure that absolutely could not go down. Multiple data streams, 24/7 operations, the pager going off at 2am. Hours spent planning outages, rehearsing failovers, trying to minimise business impact.

When I stepped away, I thought my technical skills were finished. The Java versions we’d been running were generations behind current. Frameworks came and went while we never touched them. The gap between enterprise reality and modern development felt insurmountable.

I assumed I was obsolete.

The Lightbulb

Three months ago, I started experimenting with AI coding tools. The realisation hit quickly: I can read code easily. I’m just not great at writing it.

That unlocked everything.

AI handles the syntax. I handle what matters: does this make sense? Will it break at 2am? Can I actually debug it?

Suddenly, not knowing the latest framework didn’t matter. Claude handles the syntax, the boilerplate, the “how do I do this in React again?” questions. All those operational instincts I thought were obsolete - spotting edge cases, keeping things simple, noticing when complexity is accumulating - they’re exactly what AI coding needs.

The First Debug

The moment I knew this would work was the first time I stopped copy-pasting error messages and actually debugged. Not just “make it work” but “understand why it works.”

I found myself thinking: “This can’t run out of control. If this breaks at 2am, I need to understand it.”

Those instincts I thought were obsolete? They’re the most valuable thing I bring.

What My Spidey-Sense Catches

Working with AI, I notice things:

DRY violations. The AI generates similar code in three places. It works, but AI won’t remember to go back and consolidate. That’s technical debt accumulating in real-time.

Edge cases lurking. “It’s working, but I don’t really know why” is a smell. In flight planning, that feeling meant a Christmas morning support call when our schedule was massively reduced and suddenly our assumptions broke.

Lack of resilience. Will this fail gracefully? Can I observe what’s happening? If this breaks, can I actually fix it, or will I be staring at a black box at 2am?

These aren’t framework skills. They’re operational instincts.

Back to Small Teams

The thing I didn’t expect: working solo with AI feels like working with a tight support team again.

We had 3-4 people who worked together instinctively. When incidents happened during office hours, we’d pick up roles without formal coordination:

• Someone shields and communicates
• Someone investigates
• Someone sanity-checks

No meetings about it. No Slack threads trying to coordinate across timezones. Just people who understood the system moving together.

Enterprise development is the opposite. Teams always too large. Language barriers. Timezone chaos. Communication preferences that clash. Someone’s always out of the loop, always catching up, always misaligned.

Solo + AI brings me back to that small-team clarity. I’m the communicator (via commit messages and docs). I’m the investigator (debugging what AI generated). I’m the sanity-checker (does this actually make sense?).

No communication overhead. No dependencies on other people’s availability. Just me and the work.

What Actually Transferred

The skills that matter now aren’t the ones I thought mattered:

Not important: Knowing the latest framework, keeping up with JavaScript fatigue, having opinions about build tools.

Very important:

• Framing the problem before writing any code
• Spotting where complexity is accumulating
• Asking “what breaks if this assumption is wrong?”
• Understanding what “done” means for production
• Building things you can debug when they break

My support background didn’t age badly. It aged perfectly.

I just needed the right tool to use it again.

The Irony

For years, I watched “real developers” move fast with modern frameworks while I maintained legacy systems. I assumed they had something I didn’t.

Turns out what I had - operational instinct, systems thinking, 2am debugging experience - is exactly what AI coding needs.

The frameworks don’t matter anymore. The instincts do.

And those instincts? They’re timeless.

---

If you’ve felt left behind by modern development but you understand how systems actually work in production, you might be better positioned for AI coding than you think.