I keep wondering why this topic comes up again and again. Where does this resistance to using an existing solution come from? Over the years, I’ve heard it countless times: “Why do we need that library? We’ll write our own.” I find this especially hard to understand in enterprise environments, where teams live with the consequences of these decisions for years.
Now the same idea is back with a new argument: “With AI, I can build it in a few hours.”
And you probably can. The first working version really can come together quickly. But the hardest work often starts later, when you have to discover and handle the cases nobody thought about at the beginning.
A mature library also represents years of bug reports, fixes, and tests. It carries the experience of people who have already encountered problems we don’t even know exist yet. It’s easy to compare its core functionality with our own prototype and overlook all that work.
Then you find yourself working on a project with several of these custom solutions. New features have to accommodate their limitations. Workarounds accumulate, with more workarounds built on top. Reversing the original decision becomes a costly project of its own. Sometimes it’s so difficult that the team simply stays with what it has.
Building your own solution can make sense. But “I can write it quickly” isn’t enough of a reason. AI speeds up implementation, while discovering edge cases, verifying behaviour under real conditions, and maintaining the result still take work.
It feels like a never-ending story. The tools change, but the temptation stays the same: we focus on how quickly we can get started and underestimate what we’ll have to live with afterwards.