First-order modal logic (FOML) provides a natural logical language for reasoning about modal attitudes, while retaining the richness of quantification for referring to predicates over domains. However, FOML is notoriously bad computationally, as most of the useful fragments of the logic are undecidable, over many model classes. Over the years, only a few fragments (such as the monodic fragment) have been shown to be decidable under heavy restrictions on the syntax. In this talk, I survey our recent work on the newly discovered bundled fragments based on constructions bundling quantifiers and modalities together. The idea came from our earlier work on epistemic logics of know-how/why/what, and it led us to many expressive and decidable fragments of FOML without restricting the number of variables or the arity of the predicates. I will give an almost complete picture of the (un)decidability of all the basic bundled fragments of FOML over increasing and constant domain models. I conclude with some future directions.