The problem is that you can see the foreground and depth ok as the monocular screen on my left eye 'merged' semi-transparently due to brain processing. I assume this is a bit worse on the XReal. The main issue was that when walking you make a slight sinus wave up and down compared to the foreground. You don't notice this usually, but with a paragraph of text or code positioned in front of your eyes it becomes very distracting.
One solution, using a mode of transport that doesn't involve moving up and down slightly, for example using a bicycle or car for transportation. In both circumstances, the latter being the most problematic, it's not advisable safety wise (or even illegal) and a screenreader solution is better. I had the idea of using Emacsspeak for this, or do a smart speakup echo from the commandline.
Another solution is using RSVP, or using both RSVP and text to speech. Samsung DeX is great though, Motorola, Huawai and recent Android have support for desktop mode too (if the phone supports video-out).