Tagged reranking · show all articles
Retrieval is sold as a cheap fix for a model that does not know your data. Embedding compute, vector storage, ANN queries, reranking, and index freshness are five separate meters running underneath it, and none of them appear in the price of the model call.
Retrieval-augmented generation fails quietly. Ten distinct, documented mechanisms, from a stale index to a fabricated citation, account for nearly every expensive production RAG incident, and each has its own detection and its own fix.
Retrieval-augmented generation is usually described as one step: ask, then answer. In production it is six separate stages, each with its own literature, its own failure mode, and its own knob to turn.
A production retrieval system is not one architecture but six coupled engineering decisions — chunking, hybrid weighting, reranking, access control, index freshness, and what gets watched once it is live.