What is RAG (Retrieval-Augmented Generation)?

RAGAIKnowledge BaseData RetrievalLLM

RAG (Retrieval-Augmented Generation)

RAG is a method that allows a model to access external data sources (like company documents, recent news, or a distinct database) during generation to produce answers.

Why is it important?

Standard LLMs are limited to their training data, which has a cutoff date. RAG bridges this gap by:

  1. Retrieving relevant information from a live knowledge base.
  2. Augmenting the user's prompt with this retrieved context.
  3. Generating an answer based on both the question and the retrieved data.

Industrial Use Case

Imagine an AI assistant for a Cloud & IoT Platform. Using RAG, the assistant could look up the specific, real-time status of a connected device or consult the latest user manual to help an operator troubleshoot a specific error code, without the AI model itself needing to be retrained.

Technical Terms Glossary Back to