Retrieval-Augmented GenerationRAG
Giving the AI access to your documents before it answers.
RAG combines a search step with a generation step. Instead of relying solely on training data, the model first retrieves relevant passages from an external knowledge base, then generates an answer grounded in those passages.
Pipeline: user question → embed query → vector search over your documents → top-K relevant chunks → insert into context → LLM generates answer.
This solves two core LLM limitations: knowledge cutoffs (training data ends at a date) and hallucination (the model can now cite real sources). It's the backbone of most enterprise AI search and customer-support systems.
In plain terms
Like an open-book exam instead of a closed one — the model looks things up in your documents rather than guessing from memory.