Definition: RAG

(Retrieval-Augmented Generation) An AI system that is augmented by the retrieval of data from external sources. For example, a chatbot may use RAG to search the Web for additional information. An AI search engine uses RAG to search a database of links while using its large language models (LLMs) to provide a direct answer to the query. Another RAG example is a chatbot for customer service that uses the LLM for natural language interaction and an external database for product specifics.

Agents Use RAG
Although AI agents can access external sources, agents access data for many purposes, and they are more autonomous than a RAG model. See AI agent, chatbot and large language model.