Hard Truths

The last week was one of the hardest I’ve had in the past couple of years. To explain why, I need to start a bit earlier.

Edit: This blog post was written in a moment of doubt. A couple days and 20+ hours of Claude Code usage later I am feeling more positive regarding this change already.

My career so far

A few years ago, I was working as an engineer in R&D in Germany.
It was a good job by any reasonable measure: stable, respected, and with a very promising career path. I was good at it too, which made what I did next harder to explain.

I quit that job to learn how to code.

I started with The Odin Project and got hooked fast. All I wanted to do was write code. For me, the fun part wasn't creating something out of nothing or shipping a product. It was the craft itself. I know that sounds strange to some people, but to me it felt like art: a way to express yourself, just with more logic involved.

I spent seven months learning and eventually got hired as a backend developer. For the past two years, I've been doing exactly what I love. It felt like proof that following my passion was the right decision, despite learning everything by myself, living off my savings, counting almost every Baht, and going from a high-paying engineering job back to an intern salary. Even now, I'm still far from what I earned back then.

A growing unease

From the moment I joined the company as an intern, I was placed in the AI department. Because of that, I followed the development of large language models with both interest and a bit of anxiety.

When GPT-3.5 came out, people immediately started claiming that coding, or even software development as a job, would become obsolete. Since I worked with these models every day, I wanted to see for myself how capable they really were. I tested many of them and felt that we were still far away from letting them reliably build working software. That calmed me a little.

But month after month, new models came out, and progress was fast. Social media exploded with "vibe coding" and bold claims that coding was dead. I kept testing models and remained unconvinced, though I had to admit they were improving quickly.

To cope, I mostly stopped using social media. At the same time, I noticed myself becoming more and more pessimistic about AI models. In hindsight, that was probably a defense mechanism. Coding had become my entire identity. I had barely done anything else for two and a half years.

So whenever someone at work said, "Yeah, this model is good, I use it for coding," I instinctively tried to find arguments against it. Not because those arguments were always strong, but because I needed to believe it wasn't a threat to my job, or to who I thought I was.

Then last week, something happened that shook that belief.

A wake-up call

I noticed a lot of activity in our company’s Git repositories and a new project with a very obvious name. I also heard our boss, an AI enthusiast, talking frequently about Claude Code.

Out of curiosity, I read through the repository. Around the same time, I came across a LinkedIn post from my boss outlining his vision for the company: everyone using Claude Code, with software developers mainly reviewing code, guiding agents, and steering the process.

We also had our annual kickoff coming up, and I started connecting the dots.

What I felt when I realized what this might mean is hard to describe. It wasn't just fear that my dream job could turn into something more generic, where the part I enjoyed most, actually writing code, would shrink. It was also the feeling that something I had built my identity around was slowly being replaced.

I went straight into what I can only describe as an existential crisis. I know this might sound dramatic, but for me it felt very real.

Facing my fear

Over the weekend, I tried to calm down, but my thoughts kept returning to the same questions. I slept poorly and kept imagining what my job might look like in a few months.

At some point, I decided I couldn't keep avoiding it. If this was coming, I needed to face it. So I bought the Pro plan and installed Claude Code on my machine.

What followed was hard to accept.

Claude Code still makes a lot of mistakes. Even with careful prompting, it needs constant guidance to produce high-quality code. You can't just tell it to build something and walk away. It hallucinates, makes wrong assumptions, and sometimes breaks things that were working fine. I found myself constantly correcting it and steering it back on track.

But it's also genuinely useful.

At work, I now have access to the Max plan, and I can already see advantages in almost every task. It clearly increases my efficiency. Some people talk about 10x productivity gains. For me, it feels closer to 1.5x or 2x right now. Most problems I encounter at work aren't new. They've been solved thousands of times before, and AI is often faster than I am at reproducing those solutions.

Facing the truth

I have to admit that I was wrong about AI models. Not completely wrong, but overly critical, mostly because I was scared.

That fear made me dismiss anything that threatened what I had built my identity around. Every time someone praised a model, I searched for reasons to tear it down. That wasn't skepticism. It was defensiveness.

AI will probably never fully replace software developers, but the job is going to change, whether I like it or not. I don't know yet if I'll enjoy this new version of the work. I need time to adapt and see whether the craft I fell in love with still exists in some form.

Maybe the joy of solving problems and writing elegant logic remains, even if the tools change. Or maybe it becomes something else entirely. I honestly don’t know yet.

I'll stay picky and critical of these models, especially while they keep making mistakes. That's part of who I am, and I think it's valuable. But I also need to learn to accept change instead of fighting it out of fear.

Sometimes it’s hard to accept change, especially when it affects both your job and your sense of identity. I spent two and a half years building myself around one thing, and now that thing is shifting beneath me.

In the end, though, I have to live with it. The world doesn’t care about what I want it to be. I’d rather make the best of what’s happening than ignore it and risk getting left behind.

So here I am, figuring it out.