Glossary

What is RAG (Retrieval-Augmented Generation)?

Retrieval-Augmented Generation (RAG) is a technique that improves an AI model’s answers by retrieving relevant information from an external knowledge source at query time and feeding it to the model as context, instead of relying only on what the model learned during training. A typical RAG flow searches a document store or vector database for the most relevant passages, then asks the model to generate an answer grounded in those passages.

RAG reduces hallucination and lets models answer from current, domain-specific data without retraining. Its quality depends entirely on the retrieved data. Stale, duplicated, or poorly structured sources produce poor answers. CUBIG’s LLM Capsule is a context-preserving data layer for AI that runs RAG and agent workflows on controlled enterprise context, so sensitive enterprise data can ground model output under your own context controls.

Frequently asked questions

What is Retrieval-Augmented Generation (RAG)?

RAG improves AI answers by retrieving relevant information from an external source at query time and giving it to the model as context, rather than relying only on training data.

Does RAG stop hallucinations?

RAG reduces hallucination by grounding answers in retrieved data, but accuracy depends on the quality and freshness of the source it retrieves from.

RAG vs fine-tuning, which do I need?

Fine-tuning changes the model's weights for style or domain. RAG keeps the model fixed and supplies fresh context at query time. Many systems use both.