also: retrieval-augmented generation
What is RAG?
Last updated · 1 min read
Two words, one idea: retrieval (find the relevant passages) + generation (write the answer from them). It is the standard fix for an LLM not knowing your private or recent information.
You meet RAG whenever a chatbot answers "from the help center", when Perplexity cites web pages, or when ChatGPT searches before answering. Full walkthrough: What is RAG?