RAG (Retrieval-Augmented Generation)
RAG is how you give a model access to your specific, current information instead of relying on what it memorized during training. The system first searches (retrieves) relevant chunks of your documents/data, then hands those chunks to the model along with your question, so it answers grounded in real material instead of guessing.
Practically: your documents get broken into chunks, converted into "embeddings" (see C5), and stored in a searchable index. When you ask a question, the system finds the most relevant chunks, stuffs them into the prompt as context, and the model generates an answer using that inserted text. This is the core mechanism behind popular products like Google's NotebookLM and Perplexity's search-grounded answers, and behind most internal "chat with your PDF" company assistants; developers usually wire it up with a framework like LangChain or LlamaIndex rather than from scratch.
Case example: think about ProvaCV or AfzalOS - any tool where an AI needs to answer using specific documents (a CV, a vault of notes) rather than general knowledge is a RAG-shaped problem, even if it's built simply.
What core problem does RAG (Retrieval-Augmented Generation) solve?