Why it matters
LLMs are the “search engines” of AI search. Understanding how they generate responses — blending training data with optional live retrieval, exhibiting non-determinism, and preferring certain source types — is a prerequisite for effective AEO/GEO work. Each LLM has different behavior: ChatGPT uses Bing for retrieval, Perplexity always searches live, Claude relies more on training data.How Cited uses it
Cited tracks LLMs from OpenAI, Anthropic, Perplexity, Google, and xAI, each queried with the same set of consumer-authentic prompts. The full tracked set is documented in Which AI platforms we track. Each LLM contributes independently to a brand’s overall visibility picture.Related concepts
- Which LLMs we track — the specific models and rationale
- Non-determinism — a fundamental property of LLMs
- RAG — how some LLMs augment training data with live retrieval