Even in early stages of development Capture & Replay offers a easy way to create tests. An early start of such regression tests and the assignment of unique names / IDs are the most important bases for efficient (due to stable recognition) test automation.
Sometimes it is necessary to perform mathematical calculations in QF-Test, e.g. in order to calculate the expected height of a GUI element in dependence of the complete window size or the exact clicking position for a mouse click.
Software bugs are annoying - for users and software manufacturers alike. The latter, however, should use bugs to continually improve the software, preferably from the beginning of development. Nevertheless, there will probably never really be flawless, right? Rather a philosophical question.