"One weird trick" is still pretty much the first way to go for most recommendation systems that are large embeddings focused. see torchrec, nvidia's hugectr
the thing is humans have most efficiently encoded (in detail) reality in text. humans already highlight what is worth encoding about reality.
for example, you can finetune gpt-2 to have an idea of sexual biology by having it read erotica. just like how you can have a model learn the same by watching porn. but it is much more efficient to read the text, since there is much less information that is "useless"
google deployment manager. it has incredibly subpar support for other GCP services. all support interactions resort to them suggesting migration to Terraform. we do use it in production, but not without great headache.
this isn't an explicit kill-off, but certainly purposefully offering bad support