What is retrieval-augmented generation (RAG)?
Retrieval-augmented generation, or RAG, is a way of making an AI model answer from specific documents. Before replying, the system searches your content for the passages that match the question and hands them to the model, which writes its answer from those passages.
Also called: RAG, retrieval augmented generation.
By Santhul Joseph, AI EngineerUpdated
How RAG works, step by step
- Your content (web pages, PDFs, documents, Q&As) is split into short passages.
- Each passage is indexed so it can be found by meaning, not only by exact words.
- When a customer asks a question, the system finds the passages that answer it.
- The model writes a reply using those passages, in the customer's own language.
Why it matters for customer support
A language model on its own only knows what it learned in training. It doesn't know your opening hours, your return policy or this month's prices. RAG gives it your facts at the moment it answers, so replies match your business and change as soon as your content changes.
RAG in SimplyBoost
Every SimplyBoost agent works this way. You add your website, PDFs (scanned ones too), Word documents, CSV files, text snippets and Q&A pairs, and your Shopify store's policies and pages if you sell there. Files can be up to 50 MB each. Questions the agent could not answer from your knowledge are collected under Missing answers, so you can see exactly what to add.
Questions people ask
Is RAG the same as fine-tuning?
No. Fine-tuning changes the model itself through extra training. RAG leaves the model as it is and gives it your documents at the moment it answers. For support, RAG is easier to keep current: change a document and the next answer uses it.
How do I update what the agent knows?
In SimplyBoost, edit or add a source in the Knowledge Base. Text and Q&As are ready within about a minute. After a bigger change to your website, add the changed pages again.
Do I need technical skills to use RAG?
Not with SimplyBoost. You add a website address or upload files, and the splitting, indexing and searching happen for you.