Here is what I was thinking: The point (-2,3,1) should be transformed into spherical form. Then, vector A would be transformed using the matrix. As the matrix contains cos(theta) and cos(phi), those theta and phi values would be used from the transformation of the point.
Does this make any sense?