Retrieval-augmented generation

abbreviation RAG

retrieval-augmented generation

Retrieval-augmented generation (RAG) is a technique that combines a language model with a database of external information. Instead of relying solely on pre-trained knowledge, RAG first searches for relevant documents or data before generating a response.

It works by using a retriever component to pull up the most relevant information from a knowledge base, then passing that to a generator to craft a human-like answer. This approach is especially useful for answering questions about up-to-date or specialized topics.

For professionals, RAG changes how they interact with AI tools. It reduces errors by grounding responses in verified data, but it also requires access to well-maintained databases. For example, a customer support agent using RAG might get more accurate answers about product features, but the system’s effectiveness depends on the quality of the company’s internal documentation.

A doctor using a RAG-powered medical assistant gets up-to-date treatment guidelines by pulling from the latest clinical studies.

See also: prompt engineering

Découvrez nos autres services

ElpisIA, c’est tout un univers — explorez nos autres services.