Most developers who were laid off were not the key people driving the architecture and long-term success of products. If you want to be a developer, then commit to it fully. Don’t treat it as a temporary opportunity. People forget how difficult programming and engineering really are. With the rise of AI, many non-developers think coding is the easiest thing in the world—until they actually try it. The reality is the exact opposite.
After 25 years of building applications and working with clients across multiple countries, I can tell you that clients want good products. And good products cannot be created entirely through vibe coding. Sure, you can build experiments, prototypes, and impressive demos that way, but a production-ready product requires attention to detail, solid architecture, and careful engineering. You cannot achieve that without understanding how the system is designed.
AI is here to stay, and we should absolutely use it. But we need to use it correctly, otherwise it can easily backfire.
I actually tried vibe coding this product: Izubizu Slider
I didn’t start from scratch—I already had a solid foundation. But when I reached the complex logic that handles mesh transitions between slides, I got lazy and let AI generate most of it without really reviewing the code. At first, I was amazed. I thought, “Wow, I can watch a movie and let AI handle the hard parts.”
I got about 90% of the way there, and then AI got stuck on a bug. I knew exactly how to describe the problem because of my experience, but it didn’t matter. Every attempt to fix it generated more and more code bloat until the project became almost impossible to maintain. Eventually, I threw it away and started over, oh and this experimete cost me a weak of promtpng my soul that I got to a point I was so angry that I wnated to give up evrey five minutes and also costed me about 600$ yes AI is not cheap anymore!
The second time, I did it properly. I used AI for shader work and for some smaller pieces of functionality related to mesh dragging and transitions, but I carefully reviewed everything, understood it, and adapted it to fit the architecture.
The lesson is simple: a beginner will never vibe code a slider like this to the level of detail that an experienced developer can. Not because AI isn’t capable of generating code, but because beginners don’t know what they don’t know. They can’t evaluate the output, identify architectural problems, or understand the trade-offs,
So don’t give up. Keep learning. Keep improving your coding skills. Software engineering is still important, and I believe things will get better. Companies are already realizing that AI without experienced developers isn’t nearly as valuable as many people assumed.
To become a good developer, you need to write code. That doesn’t necessarily mean writing every line manually—you can absolutely use AI—but you must understand what the code is doing. If you don’t, the project will eventually become unmanageable, although I do recomend writing some of the code for practice…
If you look at many of the vibe-coded projects on YouTube, you’ll notice the same pattern. They look impressive at first glance, but once you dig deeper, they have issues everywhere. You can tell immediately that they’re missing proper architecture and engineering discipline.
It’s also interesting to see how the AI narrative keeps changing. A few months ago, people were saying AI would replace developers within six months. Now many AI CEOs are claiming it will create more jobs instead. The messaging changes constantly, but one thing remains true: building quality software still requires experienced engineers who understand systems, architecture, and long-term maintainability.
