The main issue I've found with Agile is that in the two iterations I've seen (one BigGovCorp, one late-stage startup), the team ends up with a situation analogo

The main issue I've found with Agile is that in the two iterations I've seen (one BigGovCorp, one late-stage startup), the team ends up with a situation analogous to audience members directing a movie. You can collect a list of audience requirements (explosions, romance, Chris Hemsworth, a string quartet, Garfield) and then your devs can string them all together into a single package that technically works, but there's no rhyme or reason or predictable trend of what the software is. Furthermore, as other commenters are mentioning, once you start focusing on your ticket system, developers feel "done" when the tickets are done, rather when they believe that what they've implemented is actually a functional, conceptually consistent piece of software. I don't think there's any solution except "get the right people and then try a bunch of crap and stick with whatever gets the best results measured against the goals of leadership."