At that point, I think it becomes far more of a creative exercise than most TLs, with the writer essentially painting an entirely new history. I think "different, unrecognisable, but strangely familiar" is probably what will happen. And that's not necessarily a bad thing. You just need to make the "familiar" part not be blatant parallelism of real history and make sure everything has a cause and effect.
Who would replace them? Pre-I do European gynocrats, basques?
Isn't the matriarchal tendencies of early Europe far overstated anyway? Matriarchy is a poor word to describe those sorts of societies, be they the hypothetical pre-Indo-European ones or more recent examples in Africa, parts of Asia, etc., simply because no true matriarchy exists and probably never has, according to mainstream anthropology.
For India you have plenty of choices, though. Even if southern Indian languages tend to have large amounts of Sanskrit/Indo-European loanwords (though Tamil has less).
But one of the hazards of early enough PODs is you pretty much are forced into making your own languages (or looking up long-dead ones), at least for naming places and people. There will be different movements of people, different successful/unsuccessful groups, etc. For pre-Indo-Europeans, there's the Basque and related groups, probably a Finno-Ugric group speaking a language related to Sami in most of Sweden/Norway and probably Finland too (but the Baltic Finns migrating to Finland/Estonia would be different than no Indo-European migrations). Also there's a bunch of non-Indo-European languages scattered around the Mediterranean, some well-enough attested like Etruscan. It's a shame no one's deciphered Linear A or Indus Valley script.