Brady and Watt have constructed combinatorial $K(\pi,1)$'s for the spherical Artin groups $B(W)$, relying on their dual braid presentation and the lattice structure of the set $NC(W)$ of noncrossing partitions of the associated reflection group $W$. Moreover, after Deligne's proof of Brieskorn's $K($\pi$,1)$ conjecture these models are homotopy equivalent to the space of regular orbits of $W$.
They are remarkably simple to describe, as (topological) quotients of the order complex of $NC(W)$, yet their relation to the geometry of $B(W)$ seems to lack a good explanation. In particular, the fact that the dual and standard presentations agree was until recently proven only in a case-by-case way (relying on the classification of finite Coxeter groups).
In a seminal work, Bessis related the dual presentation of $B(W)$ with the study of the braid monodromy of its discriminant hypersurface. Then, he used the combinatorics of $NC(W)$ as a recipe to construct the universal covering space of the discriminant complement and thus prove the $K(\pi,1)$ conjecture for all $\textbf{complex}$ reflection groups.
We will explain how this study of the braid monodromy of the discriminant, along with some of our recent enumerative results gives a uniform proof of the dual braid presentation for $B(W)$. More importantly, and following a suggestion of Bessis, we will describe a way to build via this geometry a CW-complex that is a priori homeomorphic to the discriminant complement but also in fact a realization of the Brady complex. This was proposed originally as a way to simplify part of Bessis' work, but seems also well-suited for the study of CAT(0) properties of $B(W)$ via Brady and McCammond's orthoschemes. This is work in progress.