Glossary / Retrieval-augmented generation
What is Retrieval-augmented generation?
Retrieval-augmented generation, or RAG, is giving an AI model your own documents to read before it answers, so the answer comes from your material.
A model on its own answers from what it learned in training, which does not include your pricing, your product or last week's support tickets. RAG fixes that by retrieving the relevant pieces of your own content first and putting them in front of the model.
The practical effect is that answers stop being generic and start being about you. It is also the main defence against invented facts, because the model is summarising a document you supplied rather than recalling an average.
It is not a complete cure. If the retrieved material is wrong or out of date, the answer confidently repeats it, which makes the quality of what you feed it the thing that matters most.
In Creaiter
Creaiter writes from your brand voice, audience and past content rather than from the internet's average, which is the same principle applied to marketing copy.