it also seems costly to deploy such a robust search backend (eg Elastic cluster, vector db, reranking ensemble, LLM for complex parsing... these are not cheap t

it also seems costly to deploy such a robust search backend (eg Elastic cluster, vector db, reranking ensemble, LLM for complex parsing... these are not cheap technologies)