Grounding (retrieval-augmented generation)
Grounding is when an AI engine retrieves current documents, usually with a web search, and writes its answer from them, citing them as sources.
Without grounding, a model answers from its training data and cannot cite anything; brand mentions then depend on what the model learned months earlier. With grounding, the answer depends on which pages the search retrieves today, which is what a website can influence.
ChatGPT with search, Perplexity, Gemini with Google Search and AI Overviews are all grounded surfaces. That is why AI visibility tracking focuses on them: the loop between publishing a page and being cited closes in days, not training cycles.
Related terms
Measure grounding (retrieval-augmented generation) for your brand
LeapScope tracks the prompts your buyers ask on the AI engines every day and reports mentions, citations and share of voice against your competitors.