I liked the theory, and have been disappointed with a few companies' implementation. Generally speaking they tried to focus on 100% accurate estimates, which proved hard given the maturity of the team.

One company did SAFE, which worked reasonably well. I loathed the quarterly PI day, but I guess it had value. After a year I felt monotony creep in. Retrospectives didn't yield that much improvement. This was broken through by an improvement kata.

Having said all that, I am currently burned out from my company trying to execute a project with changing scope and fixed non-negotiable deadline as a waterfall project. I look back favorably on most half-baked scrum practices of the past. Especially when an agile coach was present for long periods of time.