Stop building everything just because you can
I think both sides are missing the point.
The real danger isn't the technology itself; it's the sudden disappearance of friction. For years, the gap between "I have a cool idea" and "I have a functional prototype" was a massive, grueling canyon. You had to learn the framework, wrestle with deployment, handle database schemas, and find a designer. Most ideas died in that canyon, and honestly? That was a feature, not a bug. It acted as a natural filter for quality.
Now, with LLM agents and advanced prompt engineering, that canyon has been filled in with a high-speed highway. You can go from a "word vomit" brain dump to a deployed landing page, backend, and DB schema in a single afternoon.
The Rise of Agency Hell
We’ve moved through different stages of developer suffering. First, we had tutorial hell. Then we hit the side-hustle obsession. Now, we are entering what I call "Agency Hell."
When the cost of deployment drops to nearly zero (aside from your API tokens), every fleeting thought starts to feel like a mandatory project. You see someone launch a successful AI wrapper, so you feel obligated to build one. You see a micro-SaaS hitting €10k MRR, and suddenly your hobby feels like a failure.
This is a psychological trap. When everything is possible, nothing feels significant. We are losing the ability to distinguish between "I can build this" and "I should build this."
A Case Study in Product Friction
I recently went through this with a personal project. I'm a photography enthusiast and I love going on photo walks through the city. I had this idea for an app to help people plan or join photo walks—finding specific landmarks or recreating famous shots.
I sat down with an AI agent to map out the user flows and technical architecture. Within hours, I could have had a working MVP. But as I was sketching the logic, I hit a wall of realization: I actually like getting lost during my walks. The very thing the app would "solve" is the thing that makes the hobby enjoyable.
If I had just blindly followed the "build everything" mantra, I would have wasted dozens of hours and several dollars in compute power building a solution for a problem that didn't actually exist.
Finding Your North Star
The most important skill for a developer in 2024/2025 isn't learning the latest library or mastering a specific model version—it's knowing what not to build.
We need to stop treating AI as an engine for mindless production and start using it as a tool for high-level intentionality. Instead of asking "How fast can I ship this?", we need to start asking "Does this actually matter to me?"
The freedom to build anything is a superpower, but without a sense of direction, it's just a faster way to burn out. Your relationship with AI will ultimately depend on whether you use it to amplify your actual interests or just to satisfy the anxiety of a constant "build" loop.