Wat er werkelijk stukgaat wanneer RAG in productie komt
Retrieval-demo's falen op manieren die onzichtbaar blijven tot er echte documenten binnenkomen. De drie fouten die het grootste deel verklaren, en wat u zou moeten meten in plaats van op gevoel af te gaan.
Concept. Dit artikel hoort bij de artikelensectie als opmaakskelet, en is uitgesloten van zoekmachines en van de site-assistent tot er een echt stuk in staat.
Dit artikel verschijnt in het Engels. De rest van de site is vertaald, de artikelen niet, omdat een machinaal vertaald technisch betoog minder waard is dan hetzelfde betoog gelezen in de taal waarin het geschreven is.
Every retrieval system demos well. That is the problem: the demo is run against twenty clean documents by the person who chose them, and production is four thousand documents chosen by nobody, half of which are scanned, and a user who phrases the question differently than you did.
Chunking is a retrieval decision, not a preprocessing one
Splitting on a fixed token count is the default because it is the easiest thing to write, and it is wrong for most corpora. A contract clause split across two chunks retrieves as two half-answers, both of which score badly, neither of which is returned.
Embeddings do not understand your acronyms
The failure looks like the model hallucinating. It is not. The retriever returned nothing relevant and the model did what it was asked to do with what it was given.
Nobody is measuring retrieval
The single most common gap. Teams evaluate the generated answer, which is the end of a chain, and conclude the model is weak. Measure recall at the retrieval step first: if the right passage was never in the context window, no amount of prompt work will fix the answer.
If you cannot say what fraction of questions retrieved the correct source document, you do not have a RAG system. You have a search box with a language model attached to it.
This post is a placeholder shipped with the writing section. Replace the body with the real piece; the frontmatter above is the only thing the site reads.