QuackbackRetrieval-Augmented Generation (RAG)
Retrieval-augmented generation (RAG) is a way of making an AI model answer from your own content. The system first searches your sources for relevant passages, then asks the model to write an answer using only those passages, usually with citations.
What is retrieval-augmented generation?
A language model on its own answers from what it learned in training, which knows nothing about your product and can be confidently wrong. RAG adds a search step. Your help articles, docs and other sources are indexed. When a question arrives, the most relevant passages are retrieved and given to the model, which writes an answer from them.
Why it matters for support
RAG is what makes AI answers about your product possible. It also makes them checkable: an answer that cites the article it came from can be verified by the customer and corrected by your team.
Keeping answers accurate
Quality depends on retrieval. Write articles that answer one question each, with clear titles. Combine keyword and semantic search, so both exact terms and paraphrases are found. Tell the model to say it doesn't know when nothing relevant is retrieved, and treat retrieved text as data, not instructions, to guard against prompt injection.
RAG in Quackback
Ask AI in Quackback's help center and widget writes a short answer built only from your published articles, citing the ones it used, and suggests related articles when it can't answer. Help center search combines keywords and meaning. The AI agent and Copilot retrieve from your help center, feedback, changelog, status page, documents and snippets, and retrieved content passes through a prompt-injection guard.
Common questions
What is retrieval-augmented generation?
Retrieval-augmented generation (RAG) makes an AI model answer from your own content: it searches your sources for relevant passages, then writes an answer using only those passages, usually with citations.
Why do AI support answers need citations?
Citations let customers check the answer against the source, and let your team see which article produced a wrong answer so it can be fixed.
Have another question? Talk to the team.