I sympathize with the nihilism, but having built tools that do just work as described, I've found that the problem is mostly rooted in a culture obsessed with novelty. And not novelty with a point (a truly better solution comes out), but novelty for novelty's sake (something new to talk about that's "popular" in dev circles, even if it's junk).

The only way to build solid things is to start with the POV that they need to exist and be stable long-term. You have to give a shit about simplicity, stability, performance, backwards compatibility, etc. You have to teach people what you know and form professional opinions around why you do or do not go with the crowd (and stand behind them, even under pressure from those who don't "know how").

If you commit to that (and I mean commit long-term, not for a year but decades), you can build things that don't break constantly, don't require constant maintenance, and don't drive you insane.