It's not obvious, and actually it's a bit of both. As I understand it, they sequence everything in the blood, human or otherwise. I think in practice they have ways to hook out most of the human DNA first, which cuts out of a lot of the work, and then look at what's left. Much will be known viruses, bacteria, protozoans etc. But, if I remember a previous talk by Lipkin correctly, even more will be unknown viruses etc, where they can tell from the gene structure it's a virus or whatever, but it doesn't match any known virus.
These unknown viruses etc will generally be present in much smaller quantities than common ones, which is why they are unknown (also, because many of them are impossible to culture in the lab, but deep sequencing doesn't rely on culturing).
Hope this helps rather than confuses.