Term explained
What is RAG (retrieval-augmented generation)?
RAG is a way of making an AI look up relevant information, such as your documents or a website, before answering, so its answer is based on real sources.
In plain English
On its own, an AI model answers from memory: whatever it absorbed during training, which may be out of date or simply not include your information. RAG, short for retrieval-augmented generation, adds a lookup step. When you ask a question, the system first searches a set of sources (company documents, a help centre, the web), picks the most relevant passages, and gives them to the model along with your question. The model then writes its answer from those passages.
It is the difference between asking someone a question from memory and letting them check the files first, an open-book exam rather than a closed-book one.
Why it matters
RAG is how most business chatbots answer questions about a company’s own products and policies, and how AI search tools give answers with links to sources. It keeps answers up to date without retraining the model, and it cuts down hallucinations, because the model has real text to work from and you can check its sources.
An example
A staff member asks the company assistant, “How many days of parental leave do I get?” The system searches the HR handbook, finds the right section, and passes it to the model, which answers in plain English and links to the page in the handbook.
What to watch out for
RAG is only as good as what it finds. If the search pulls up the wrong or an outdated document, the answer will be confidently wrong, and the model can still misread or over-summarise a correct source. When an answer matters, click through to the source it cites and read it yourself.