What is RAG (retrieval-augmented generation)?
Retrieval-augmented generation (RAG) is a technique where relevant source passages are retrieved from a document collection and given to a language model, so its answer is grounded in those passages rather than its training data.
RAG is what separates a chatbot that knows your business from one that guesses. Documents are split into passages and indexed by meaning; each question retrieves the closest passages, and the model is instructed to answer only from them.
Why it matters
A language model on its own has no reliable knowledge of your prices, policies or products, and will fabricate plausible answers if asked. Retrieval puts the real text in front of it at the moment of answering, which is both more accurate and auditable — you can show which passage an answer came from.
Examples
- Splitting a website's pages into passages, embedding them, and retrieving the closest few for each question.
- Citing the source page alongside an answer so a customer can verify it.
- Treating a weak retrieval score as "I don't know" rather than answering anyway.
Not the same as
Fine-tuning
Fine-tuning changes the model's weights to shift its style or behaviour. RAG changes what the model sees at question time. For keeping facts current, retrieval is the right tool — you re-crawl instead of re-training.
Related
term
What is a knowledge base?
Traditionally a knowledge base is a help centre humans read. For AI support it is the source material a representative answers from — which means an ordinary website, well written, can serve as one without a separate help centre.
term
What is an embedding?
Embeddings are why an AI representative can answer "do you deliver to Berlin?" from a page that only says "we ship across the EU". Text is converted to vectors, and similar meaning lands close together.
term
What is an AI agent?
The word is used loosely. In customer support it usually means something more capable than a question-answering bot — a system that can look things up, take an action, or decide to escalate, rather than only replying.
use case
Answer customer questions from your own content
Answering questions well means answering them from something true. RubyRep retrieves the relevant passages from your site before it writes anything, and cites the page it used, so an answer can always be checked.
guide
How to train AI on your website content
"Training on your website" almost never means training a model. It means crawling your pages, splitting them into passages, indexing them by meaning, and retrieving the right ones at question time. Understanding that distinction tells you exactly what to fix when answers are wrong.
industry
AI customer support for SaaS companies
SaaS support is dominated by questions your documentation already answers: what a plan includes, whether a feature exists, how an integration works. RubyRep reads your docs, pricing and changelog and answers those on the site, so trial users get unblocked in seconds instead of waiting on a ticket.
RubyRep is this, applied to your website.
Give it your URL and it reads your pages, then answers your customers from them.
Train My RubyRep