Contact

RAG (Retrieval-Augmented Generation)

Category
Artificial intelligence
In Hungarian
RAG (visszakereséssel bővített generálás)
Entry
67 / 90

In short: What is RAG (Retrieval-Augmented Generation)?

RAG (retrieval-augmented generation) is a method in which a language model first retrieves relevant documents and then bases its answer on them.

Explanation

The process works in two steps. First, the system retrieves the passages most relevant to the user's question from a knowledge base, typically using embeddings and a vector database. Next, the language model receives those passages as context and drafts the answer.

This lets the model work with fresh, company-specific information without retraining. Sources can be cited alongside the answer, and the risk of hallucination drops significantly, though it never disappears completely.

Example: an internal assistant finds the relevant paragraphs in company policies and manuals, answers the employee's question and links to the exact source document.

All terms from A to Z

Have an idea or an engineering challenge to solve?

Let's talk