The tldr is:

(almost) No one needs vector databases when they have massive context models.

If you're a vector database vendor, you probably disagree; but you're also wrong, and biased.

You also look bad when you say things like 'RAG solves hallucinations' which is objectively false, sales pitch rubbish.