What is retrieval?
Retrieval means finding relevant information before producing an answer. Instead of relying only on a model's training, a system searches a collection of documents or another source.
How RAG works
Retrieval-augmented generation, or RAG, combines a search step with a generation step. The selected passages are supplied to the model as context for its response.
What embeddings do
An embedding represents information as a list of numbers. Similarity search uses those representations to find passages that may be related to a question. It is a useful starting point, but does not establish truth.
What matters
Good source material, sensible document boundaries and useful search results matter more than a complicated pipeline. Check that the retrieved passages actually support the answer.
Key takeaways
Retrieval gives a model relevant context. It does not remove the need to check sources, permissions, freshness or the model's final claims.