Or, X was initially composed of A alone; B appeared, (perhaps as a result of the mutation of A into two copies of A, a common occurrence, and then as B could be used to perform some of the functions of A, A and B both modified in ways that made B required. Now A and B together appear to be irreducibly complex.
This outcome of evolution was first spotted by the Nobel prize winning biologist H. J. Muller in papers that came out in 1919 and 1939. [Edit] Specifically, the mechanisms go under the name Muller's Morphs) [/Edit]
So, Behe took something previously shown to be a natural consequence of evolution and, uh, missed the point somewhat spectacularly.