I think there's some freeware kicking around that generated gain trims from reading the meter levels to make a crude NOM algorithm. The main problem with using (rectified and filtered) meter data instead of a proper sum of all channels, is it can't discriminate between single coherent sources hitting multiple mic inputs, and multiple incoherent sources (a 3dB error between optimal gain sharing for those two cases). You could probably split the difference between the two making a worst case +/- 1.5 dB error from ideal (I don't know what the freeware does, not that bad of a choice to just assume incoherent sources).
The basic gain sharing algorithm could be coded correctly inside a digital mixer with access to unfiltered data pretty easily, but Dan has decades of experience and many improvements that he is also selling in his product. I still have a hard time thinking of this as inexpensive, but compared to his previous product it surely is. I wish him much success, he deserves it for his contributions in this area.
JR