> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getcited.in/llms.txt
> Use this file to discover all available pages before exploring further.

# RAG

> Retrieval-Augmented Generation — the technique of combining LLM generation with real-time web retrieval.

**RAG** (Retrieval-Augmented Generation) is the technique of combining LLM text generation with real-time web retrieval. Instead of relying solely on training data, a RAG-enabled system searches the web for relevant sources, then uses those sources to generate a grounded response with citations.

## Why it matters

RAG is how [Perplexity](/concepts/platforms/how-perplexity-ranks-sources) and [ChatGPT](/concepts/platforms/how-chatgpt-search-works) (in search mode) produce answers that reference current web content. For brands, RAG creates an opportunity that purely training-data-based models do not: recently published content can influence AI answers within days or weeks, rather than waiting months for a training data refresh. It also means that web content quality, crawlability, and freshness directly affect whether a brand is cited.

## How it applies in practice

Cited tracks both RAG-enabled platforms (Perplexity, ChatGPT search mode) and training-data-primary platforms ([Claude](/concepts/platforms/how-claude-search-works), [Gemini](/concepts/platforms/how-gemini-search-works)). The brand visibility patterns differ between these two groups — RAG-enabled platforms respond faster to content changes but show more run-to-run variance because retrieval results change between sessions.

## Related concepts

* [How Perplexity ranks sources](/concepts/platforms/how-perplexity-ranks-sources) — the most RAG-dependent platform
* [How ChatGPT search works](/concepts/platforms/how-chatgpt-search-works) — RAG via Bing integration
* [Non-determinism](/glossary/non-determinism) — RAG adds an additional source of response variance