Retrieval-Augmented Generation (RAG)
What is Retrieval-Augmented Generation (RAG)?
Retrieval-augmented generation (RAG) is a technique in which an AI system first retrieves relevant passages from a trusted source, such as a knowledge base, and then generates its answer from those passages.
RAG is the most common way to make generative AI answer from your own policies, product information and procedures instead of from whatever the model happened to learn in training. When a customer or agent asks a question, the system searches the approved content, passes the best matches to the language model with the question, and asks it to answer only from them, ideally citing the source.
RAG reduces hallucinations and lets answers change the moment the knowledge changes, with no retraining. Its quality depends on the knowledge itself: articles that are current, owned, well structured and written for the question people actually ask. That is why knowledge management and AI projects in the contact center tend to succeed or fail together.
Related: Grounding, Contact center knowledge base, UW Knowledge & Assist.