What is retrieval-augmented generation (RAG)?

Definition: Retrieval-augmented generation (RAG) is a technique where a system first retrieves relevant information from a knowledge source and then passes it to a language model to generate an answer. The term comes from a 2020 research paper by Patrick Lewis and colleagues, "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks."

How RAG works

Documents are split into chunks, converted into embeddings and stored in a search index, often a vector database. When a user asks a question, the system finds the most relevant chunks, adds them to the prompt, and the model answers based on that context. Many systems combine vector search with keyword search and a reranking step, and return citations so users can check the source.

When a company needs it

RAG fits when answers must reflect private or frequently changing information, such as policies, contracts, product documentation or support history, without retraining a model. Quality depends heavily on chunking, retrieval, access controls and evaluation, which is where most engineering effort goes.

How Ryz helps

Ryz RAG engineers and AI pods build retrieval systems in your cloud and repos, working on your team with models from providers such as Anthropic and OpenAI.

Related

• Hire RAG engineers
• RAG development services
• RAG vs fine-tuning
• Vector database