What is Retrieval-augmented generation (RAG)?
Letting a model look up relevant documents first and answer using them.
In plain English
The system searches your own documents for passages related to the question, adds them to the prompt and asks the model to answer from them. This keeps answers current and grounded in your sources without retraining the model, and it lets you show citations. Quality depends heavily on how well the retrieval step finds the right passages.
Example
A support assistant searches the company help centre, then answers using the three most relevant articles.
Easy to confuse with
- Fine-tuning: RAG supplies knowledge at question time. Fine-tuning changes how the model behaves or writes by further training.
Where you will meet it
This idea belongs to the subject area of these practice tests. We do not publish or predict real exam questions.
- AWS AIF-C01 practice tests
- AWS AIP-C01 practice tests
- Google Cloud Generative AI Leader practice tests
- Databricks GenAI Engineer Associate practice tests
- NVIDIA NCA-GENL practice tests
Related terms
- Embeddings and vector databases: Lists of numbers that capture the meaning of text or images, stored so similar items can be found quickly.
- Hallucination: When an AI model states something false or invented as if it were true.
- Fine-tuning: Training an existing model a little more on your own examples so it behaves in the way you want.
- Tokens: The small chunks of text that a language model reads and writes, and that usage is measured in.
Test yourself
Practice tests with an explanation for every answer option, so each mistake teaches you something.
Browse practice testsAll glossary termsWritten and checked by the Be Ready Education team, last reviewed 2 November 2026. General study information, not professional advice. Providers change names and rules, so confirm details on the official documentation. Not affiliated with, endorsed by or sponsored by any certification vendor or exam body. Certification names, exam codes and product names are trademarks of their respective owners and are used only to identify the exam a practice product prepares for. Our questions are original and are not actual exam questions.