RAG
AI that looks things up in your trusted information before it answers.
RAG stands for retrieval-augmented generation. Before answering, the AI searches a trusted collection of documents, like your files or a company's help pages, pulls out the most relevant parts, and writes its answer from them. Fewer guesses, more real answers.
How it works
- 01Your question is turned into a search.
- 02The most relevant passages are retrieved from trusted documents.
- 03The AI writes its answer using those passages, ideally showing where each fact came from.
Checking the binder before you answer
Your partner asks, "When's the next check-up?" You could guess. Or you could open the folder, find the page and answer properly. RAG makes the AI do the second one: look it up in trusted documents first, then answer. That's what happens behind the scenes when you upload a file, or when a website's assistant answers from its help pages.
Where you'll use it
Your own family FAQ
Ask questions about your uploaded daycare, school or insurance documents.
Support assistants
Website chatbots that answer customers from your real FAQ and policies.
Answers with sources
AI search tools that show the links they used, so you can check them.
Your nap-time experiment
Run your own mini RAG. Upload a document, or paste a long text like a FAQ page, and ask:
Using only the document I've shared, answer these questions: 1. [your first question] 2. [your second question] 3. [your third question] For each answer, quote the short sentence from the document you based it on. If the answer isn't in the document, say "Not in the document" instead of guessing.