Free-electron lasers produce X-ray pulses with a peak brightness a billion times that of beams at a modern synchrotron radiation facility. This has provided a disruptive new technology, in several senses of the word. A single focused X-ray FEL pulse completely destroys a small protein crystal placed in the beam, but not before that pulse has passed through the sample and given rise to a diffraction pattern. This principle of diffraction before destruction has given the methodology of serial femtosecond crystallography for the determination of macromolecular structures from tiny crystals without the need for cryogenic cooling. Consequently, it is possible to carry out high-resolution diffraction studies of dynamic protein systems with time resolutions ranging from below 1 ps to milliseconds, from samples under physiological temperatures and other conditions. The high intensity and coherence of the X-ray beam can also be exploited for novel phasing approaches, ranging from preferential ionisation of elements to the use of intensity measurements between Bragg peaks. Even now, a decade after the first experiment at LCLS, we have not fully explored the limits of the technique, nor developed it to its full potential. I will discuss some of those potentials.