In plain English
RAG, short for retrieval-augmented generation, improves accuracy by connecting a language model to an external knowledge source such as a company's documents, a database, or the web. Instead of relying only on what it learned in training, the model first retrieves relevant information and then uses it to generate a grounded response. It is widely used in business AI to reduce hallucinations and keep answers current.
Why it matters
RAG lets a general model answer questions about your specific, up-to-date information without retraining it, which is why it is the default approach for most enterprise assistants.
A worked example
An internal helpdesk bot retrieves the latest HR policy document, then answers a leave question based on that text rather than on its training data.
Common confusion
RAG does not retrain the model. It supplies relevant documents at the moment of the question, so updating your knowledge base is enough to change the answers.