Well, it predicts infrared spectroscopy to train detectors accurately enough to chemical structures. It can calculate other spectroscopic information so as to identify unknown materials being investigated. I'm sure there are other microscopic properties that can be investigated, such as tunnelling or material/catalysis design, but I've only worked in spectroscopy.

Edit: forgot... Minimization to transition states has been used to understand chiral synthesis and has been proposed as a means of training antibodies to assist in such synthesis. It's also been used to humanize antibodies (in combination with the classical forcefield molecular dynamics).

Are these all "simple experiments". Yeah, I guess so. No extreme physical conditions are required to perform them. Can be done in high school labs, ... Still, they're specific calculations of physical properties that reproduce nature. Not sure your point re: "doesn't describe reality" - Can relativity predict ir spectroscopy (not the shift thereof due to acceleration)? Does string theory predict anything testable vs. simpler theories?