What is RAG?
Retrieval-Augmented Generation: an architecture pattern that grounds LLM responses in real, retrieved documents rather than relying solely on the model's training data. A RAG system first searches a knowledge base (documents, databases, APIs) for relevant context, then feeds that context to the model alongside the user's question, which sharply reduces hallucinations and enables domain-specific accuracy.
My perspective
In practice
The go-to pattern for enterprise AI applications where accuracy matters. Instead of fine-tuning a model on proprietary data (expensive, brittle), RAG keeps the knowledge base separate and searchable. Common implementations use vector databases (Pinecone, Weaviate, pgvector) for semantic search. The quality of retrieval (what documents you find and how you chunk them) matters as much as the model itself.