ITG NowHome › AI glossary › RAG
What is RAG?
RAG is a method where an AI looks up facts in your documents first, then uses them to write its answer.
- Category Agents & tools
- Also called retrieval-augmented generation, retrieval augmented generation
- Updated 20 Oct 2026
RAG, explained simply
RAG stands for retrieval-augmented generation. It means the AI searches a set of documents, picks the most useful parts, and reads them before it answers.
It matters because an AI on its own can guess or miss things it never learned. RAG lets it answer from your own files.
For example, you could load a company handbook into a RAG tool. Then you ask, "How many vacation days do I get?" and it finds the right page.
RAG often uses a vector database to find matching text quickly.
On ITG Now, our free tool guides cover open source RAG projects, and our weekly lists show which ones people are starring.
Related terms
Vector databaseEmbeddingsSemantic searchHallucinationContext windowLLM
Questions people ask about RAG
RAG vs fine-tuning: what is the difference?
RAG hands the AI documents to read when it answers. Fine-tuning changes the model itself through extra training.
Does RAG stop hallucinations?
It helps, because the AI works from real text. It can still make mistakes, so check key facts.
Do I need RAG to chat with my own files?
Most tools that let you chat with your files use RAG behind the scenes.