I'm not even gonna pretend I could've come up with this idea.
This is the paper.
What did they do? They purified proteins that they are interested in that interact. They purified these from cell lines where they made mutations that changed single amino acids in the interacting sites of the proteins. Then they did native mass spec (in a modified(?) Q Exactive Plus system. It's a little unclear, it may have the vendor's native mass options "Biopharma mode" but I didn't check the reference.
If you have two proteins that interact and you can get a mass spectrum from the native configuration of those two proteins, but you control the mutations in the proteins -- you get a super cool readout! What is the strength of the interaction of those two proteins -- and, more importantly, how does it differ when you change this amino acid over here, vs this one over there?
Right? See how cool that is? How easy would it to be to extend this to your protein(s) of interest that you already had somebody construct all those mutants of?!? Sure, the math looks a little daunting at first, but they explain it in such clear detail (good writing!) that I feel like you just plug this all into Excel and only use that scary math thing at the top of this post in presentations to impress people in your department.