RAG
Retrieval-Augmented Generation end to end — and the same question answered with retrieval switched off, so you can watch the model confidently make things up.
RAG grounds a model in your documents: embed the question, retrieve the most similar chunks from a vector database, and paste them — literally, as text — into the prompt above the question. The model answers from what it was just shown instead of from its training data.
This workspace runs the real retrieval (same corpus and cosine ranking as the Vector Databases topic), shows the assembled augmented prompt verbatim, and colour-codes which retrieved chunk each part of the answer came from. The signature moment is the RAG OFF toggle: the same question, no retrieval — and a fluent, confident, wrong answer.