RAG (retrieval-augmented generation)
RAG means the model looks up documents at answer time instead of relying only on what it memorized in training.
Without retrieval, the model is a well-spoken archive with a cutoff date. With RAG, it searches an index — the web, a vector store, your Notion — then writes with those passages in context.
Perplexity is RAG-forward. ChatGPT Search is RAG when browsing is on. Google AI Overviews mix retrieval with the search index you already know. GEO is mostly a fight to be in that retrieved set.
Chunk quality matters. If your page is one blob of marketing, the retriever pulls a useless slice. If each section stands alone, you get passage-level citations.
Reddit RAG is just RAG over threads. The chunk might be a comment, not a page. That is why a 40-upvote anecdote can outrank your white paper.
Examples
- A support bot that answers from your docs is RAG. If it ignores the docs and invents a refund policy, retrieval failed or was never wired.
- ChatGPT cites a 2024 blog post for a 2026 pricing question because the live fetch missed your /pricing update. Stale RAG, not “the algorithm hates you.”
Related terms
Vectors that represent meaning. They power semantic search and RAG. Your page becomes a point in space.
Combining keyword (BM25) and vector retrieval. Many RAG stacks do this. Your nouns and your meaning both count.
The date after which the model’s weights don’t include the web. Anything newer needs browsing or RAG.
Query fan-out is when one user question becomes many internal searches. You might get cited for a sub-query you never targeted.
Finding nearest neighbors in embedding space. Core of RAG. Mushy pages cluster together and lose.
FAQ
Is RAG the same as fine-tuning? +
No. Fine-tuning changes weights. RAG leaves the model alone and stuffs evidence into the prompt. Most AI search you feel as a user is RAG plus memory.
How do I optimize for RAG? +
Clear headings, factual chunks, unique data, crawlable HTML. Then earn places in the indexes those systems search — including Reddit.
Track this in Reddex
See Reddit threads and AI answers for your brand in one place. Start with a free analysis.
Get started