I am struggling with this as well.
I am hoping maybe cpq’s idea to use the SWD pins on the underside of the XIAO to program it directly can help.
I have a chip that has become completely unresponsive (lsusb in Linux won’t see it), after it became bricked testing the latest version of bossac to upload firmware to it.
(NOTE: I probably messed up the command line arguments to bossac, without realizing how seriously it could impact the chip beyond the point of recovery with my available hardware).
I need something considerably more reliable than what is officially documented by them if I am to use this industrially. The chip seems really nice, but this really holds it back for me.