AI Gave Me a Mid-Life Crisis

I have been enjoying software development ever since my university days. Building things, thinking about edge-case scenarios, implementing guard rails against them, and automating tedious grunt work just scratches an itch in my brain that I can never really explain. For the longest time, this is how I thought software development worked. Before AI Here's the breakdown of how I was building software before AI: Discover what needs to happen. Whiteboard the high-level process. Create a blueprint of what the architecture should look like. Start building the architecture. Do the coding. Test. Realize something is broken. Scour the internet for a solution. Go back to step 5. When do you actually ship the software? When you realize things are just broken enough to be usable. When do you actually complete building the software? You don't. And honestly, I enjoyed that. What's enjoyable for me in this process of building is discovering things along the way that I didn't even know existed. You learn strategies that get engraved so deeply into your brain that, on your next project, the guard rails for those problems just come naturally. You start developing a mental list of things to watch out for and eventually, that list gets long enough for someone to call you a senior developer. At least, that's how I thought about it. And there's just this euphoric feeling when you finally find that 10-upvote Stack Overflow answer for the exact same problem you've been dealing with for the last 3–4 hours. Then AI Came When AI was first emerging, my old boss told me that developers' jobs would be gone in a few years — or at least look very different from what they looked like at the time. I was skeptical, of course. As a human being, it's unsettling to hear that eventually your entire career and experience — things you've built brick by brick — might not matter anymore because some advanced version of ChatGPT could take your steps 1 and 2, figure out steps 3 onward, and build everything 100x faster than you could. It wasn't like that at first. I tried using AI, but it wasn't particularly good. Seeing how often it hallucinated requirements and invented scenarios that could never exist only solidified my belief in the: "Developers can't be replaced by AI." Then an opportunity came to me, a software development role, the function is only building and developing software and I really like that Idea. So I quit my old job. I was kind of an all-arounder there (company culture was a bit like a start-up) and I joined to this new company that I didn't know, at the time, was going to be extremely aggressive about leveraging AI. Fast forward almost a year. I haven't manually written code in the last six months or so. We were heavily encouraged to use AI, the productivity expected from you will force you leverage AI to meet the deadlines. My Job Now My workflow looks something like this: Explain what I want to AI in Plan Mode. /grill-me /grill-me again. /grill-me again. Keep going until I'm satisfied with the plan. Let the agent build it. Test it. Fix whatever the agent got wrong. Ship it. Even my PR comments and test scenarios are AI-written. There's just no point in trying to do something manually when AI can do it much more efficiently and in a fraction of the time. And that's where my mixed feelings start. The Good Part On the upside, we ship changes incredibly fast. We build incredibly fast. It's almost like Inception. You think it, and AI builds it. Have an idea? Build it. Want to refactor something? Do it. Want to try a completely different approach? Why not? The barrier between "I have an idea" and "I have working software" has become incredibly thin. And that's genuinely fun. The Part I Don't Like The downside is that steps 3–9 were where most of the learning happened for me. That was the part where I struggled. Where I searched. Where I read documentation. Where I tried something, watched it break, figured out why it broke, and eventually learned something that stuck with me. Now? I just ask AI. And it gives me an answer. Sometimes a very good answer. Sometimes the wrong answer. But even when it's wrong, I'm usually not spending hours digging through the problem the way I used to. So I don't learn the same way anymore. And I haven't really figured out how I feel about that. Work Feels Easier and Harder at the Same Time The transition hasn't been easy. Work feels easier and harder at the same time. The actual implementation is easier. The mental workload is harder. Because when the implementation cost becomes almost negligible, the bottleneck moves somewhere else. Ideas. Requirements. Edge cases. Architecture. Resilience. "What happens if this goes wrong?" "What happens if the user does this?" "What happens if this service is down?" "What happens if this gets called 100,000 times?" The coding isn't the hard part anymore. Thinking about everything that could possibly go wrong is. And that takes a lot of mental energy. There isn't really any downtime in building anymore. As long as you have the idea, AI can build it for you. I Realized Something About Myself I enjoy the process as much as the result. Maybe even more than the result. I enjoyed discovering things. I enjoyed getting stuck. I enjoyed finding that obscure blog post or Stack Overflow answer after hours of searching. I enjoyed finally understanding why something wasn't working. AI killed a lot of that for me. But at the same time, it made building things incredibly accessible. I can go from: "I have an idea." to "It's running locally." in a ridiculously short amount of time. And that's fun in its own way. So... Where Does That Leave Me? I'm writing this while genuinely reflecting on whether this career path is still something I want to pursue. Is Plan Mode + /grill-me going to be my workflow for the next 10–20 years? I honestly don't have the answer. And I don't expect that answer to come anytime soon. I also don't have advice for anyone who is going through the same thing. I'm still figuring it out myself. Maybe this is simply what software development is becoming. Maybe the skills that made a great developer 10 years ago aren't the same skills that will make one great 10 years from now. Maybe the ability to write code was never the thing that mattered most in the first place. Or maybe I'm just going through the weirdest transition in my career so far. I don't know. I guess only time will tell.