The Conversation That Changed My View on AI

I almost didn't ask the question. I'd been carrying it around for months — quietly, the way you carry a worry you don't want to say out loud because saying it makes it more real.

We were at our annual company conference at the Marriott. Dinner was winding down, people were mingling, and I found myself standing near my manager with a glass in hand and a thought I couldn't shake anymore. So I asked.

// the moment

Annual conference. Marriott Hotel. The kind of evening where the conversations that actually matter happen between the scheduled sessions — in corridors, over coffee, between colleagues who've let their guard down a little.

I pulled my manager aside. I said: "I need to ask you something I've been thinking about for a while."

The Fear I Couldn't Stop Thinking About

Over the past two years, I had quietly watched AI start doing things I had spent a decade learning to do with my own hands. And it wasn't small things — it was the core of how I work.

🐧

Linux commands — muscle memory built over years, now typed by AI in seconds

☸️

Kubernetes YAML — I used to write it from memory. Now I describe what I need and AI writes it

⚙️

Ansible playbooks — hours of careful role structuring, now generated in one prompt

🏗️

Terraform code — modules I used to craft manually, now drafted by AI faster than I can open a file

I had worked hard for 10 years to build these skills. Not just "know about them" — but to have them in my fingertips. The kind of knowledge where you don't think, you just type. That took years of late nights, real incidents, hard-won experience.

And now I was looking at AI tools doing the same thing in seconds. And worse — I was using them. I was asking AI to write the code I used to write myself. Every time I did, a small voice in the back of my head asked: am I losing something? Is this making me weaker?

So at that dinner, I finally asked my manager directly:

"Will AI replace the actual skillset I've spent 10 years building? I'm relying on it to write code I used to write by hand. Am I losing my real skills — or am I just being paranoid?"

What He Said

My manager listened. He didn't dismiss the concern. He thought for a moment, and then he said something I've been thinking about every day since.

"Use AI as a tool. Do not 100% rely on it. Think of yourself as an architect — you're not replaced by your tools, you're the one giving them direction. You are the one who commands AI to do your task. That is the role. But always check the work yourself afterwards. Cross-check manually when you have time. Understand what AI has suggested, and ask yourself — what more can be improved? That's where your 10 years still lives. In that question."

Naveed Sanghera

Manager · Annual Conference, Marriott Hotel

Why That Answer Hit Differently

I've read a lot of takes on AI and careers. Most of them are either dismissive ("AI will never replace real engineers") or alarmist ("everything will be automated in two years"). Both feel dishonest.

What Naveed Sanghera said was different because it was grounded. It wasn't trying to comfort me — it was reframing the question.

He didn't say "your skills are safe." He said: your role is changing, and the way you engage with AI determines whether that change makes you stronger or weaker.

The architect analogy is the one that stayed with me. An architect doesn't lose their expertise because they use AutoCAD instead of a drafting table. The software doesn't replace the architect's judgment about structure, safety, aesthetics, and context. It just removes the mechanical barrier between thought and drawing.

AI is doing the same thing to my workflow. The Linux commands I used to type by hand — I still need to know what they do, why they exist, and when to use them. AI just removes the barrier of remembering exact syntax under pressure at 2am.

The real skill was never memorising kubectl flags. The real skill is knowing which flags matter, in which situation, for which reason. AI can generate the command. Only experience knows when to run it — and when not to.

How I Actually Changed After That Conversation

I stopped feeling guilty about using AI. I started using it more deliberately. Here's what that looks like in practice now:

I use AI for the first draft, not the final answer

When I need a Kubernetes manifest or a Terraform module, I let AI write the first draft. Then I read every line. I ask: does this make sense for our specific setup? Is there a security issue here? Could this fail in a way the AI hasn't considered? That review is where my 10 years is still doing its job.

I always verify manually — especially under pressure

This is the piece of Naveed Sanghera's advice I follow most strictly. AI gets things wrong. It confidently suggests configurations that work in general but break in our specific environment. The manual check isn't optional — it's the job.

I treat "what could be improved?" as a non-negotiable question

AI gives you the adequate solution. Experience helps you find the better one. After every AI-generated piece of code or configuration, I ask: what is this missing? What edge case hasn't it considered? What would I do differently? That's the question where your hard-earned knowledge still lives — and always will.

I test business logic myself, even when AI writes the tests

AI can write test cases. It doesn't know our business logic — only we do. I still define what "correct" looks like. I still catch the cases that AI missed because it didn't understand the context. The judgment layer is irreplaceable.

The Shift in How I See It Now

Before that conversation, I was thinking about AI as something happening to me — a force that was slowly eroding the value of what I knew. After that conversation, I started thinking about AI as something I could direct — a force multiplier that makes my judgment more impactful, not less.

The engineer who uses AI well is not the one who asks the best prompts. It's the one who can evaluate what comes back with trained eyes, improve on it with real experience, and take accountability for the final result.

That's still a deeply human skill. And 10 years of hard work is exactly what makes someone good at it.

    // What I Took Away From That Conversation
    AI is a tool. You are the architect. Tools don't replace architects — they extend what architects can build.
Do not rely on AI 100%. Always verify, always cross-check, especially for anything production-critical.
Your 10 years doesn't live in the commands you've memorised. It lives in your judgment about when, why, and how to use them.
Ask "what could be improved?" after every AI output. That question is where your experience still creates irreplaceable value.
Business logic is yours. Testing frameworks can be AI's. The understanding of what correct looks like belongs to you.

  

A Note to Anyone Feeling the Same

If you've been carrying this worry quietly — the feeling that AI is slowly replacing something you worked hard to build — I want you to know that I get it. It's not paranoia. The concern is real and the change is real.

But the answer isn't to avoid AI or to surrender to it. The answer is to stay in the role of the person who directs it, evaluates it, and takes responsibility for what it produces.

You are not the person who types commands. You are the person who knows which commands matter, and why, and what happens when they go wrong. AI cannot learn that from a training dataset. It has to be lived.

And you've been living it. That's still worth something. It's worth more now, not less — because it's rarer.

— Naveed Ahmed, Lead DevOps Engineer @ DigitalOcean
With gratitude to Naveed Sanghera for a conversation that reframed everything.