RAG (retrieval-augmented generation)
RAG (retrieval-augmented generation) is a technique where AI first searches your documents for relevant passages and only then writes the answer based on them. It reduces made-up answers and lets the AI cite its sources.
Documents are split into short passages and indexed. When a user asks a question, the system finds the most relevant passages and passes them to the language model together with the question.
The quality of the answers depends mostly on the quality of the documents and the search. That is why RAG projects start by cleaning up and structuring the source data. See how we used it in AI assistants for 2N and FlexiFin.
What it means for your business
RAG is the usual way to build an AI assistant over manuals, contracts or a knowledge base without training your own model.