We get asked fairly often why we tend to recommend Postgres over the newer database of the month, or a plain deploy script over a fashionable orchestration layer. The honest answer: boring technology has already had its surprising failure modes discovered by someone else.
This isn't a blanket rule. Newer tools sometimes really are the right call, especially when they solve a problem the boring option genuinely can't. But the bar for a small team taking on something new should be higher than "it's popular right now," because the cost of an unfamiliar failure mode at 2am is real and it's paid by whoever's on call.
Our rule of thumb: default to boring, and justify the exception in writing. If you can't articulate what specific problem the new thing solves that the old thing doesn't, that's usually the answer.