# Retrieval-Augmented Generation Retrieval-augmented generation (RAG) retrieves relevant external records and supplies them to a model so an answer or artifact is grounded in a corpus rather than model parameters alone. Bryant's systems supply unusually strong source material for RAG because the documents preserve paths, frontmatter, explicit links, semantic annotations, source relations, and derived identifiers. A careful implementation should retrieve canonical sources and keep generated pages as projections rather than competing truths. Related: [[wiki/Semantic Retrieval|Semantic Retrieval]], [[wiki/Metadata Grounding|Metadata Grounding]], [[wiki/Knowledge Graph|Knowledge Graph]], [[wiki/Provenance|Provenance]]. Sources: [OpenAI — File search](https://platform.openai.com/docs/guides/tools-file-search), [OpenAI — Vector stores](https://platform.openai.com/docs/api-reference/vector-stores)