Generative search engines synthesize information by retrieving, re--ranking and producing new content rather than simply listing hyperlinks. This shift challenges conventional visibility metrics: retrieved sources may not survive into the final answer, and user clicks decline when AI summaries appear. We document an observational single--case study of concept--reference emergence under generative search, focusing on GhostDrift Mathematical Institute (GMI). Despite disadvantageous initial conditions---recent incorporation, low public recognition, no peer--reviewed authority, limited human traffic and no mainstream coverage---GMI reported that its definitional pages were observed among referenced sources in generative search outputs and external media. We reconstruct a timeline of concept definitions, external mentions and AI--related observations; classify evidence into official documentation, third--party sources, self--published logs and author interpretation; and outline inclusion and exclusion criteria. Results show that AI--related queries far exceeded human sessions, that definitional pages were available prior to external citations and AI references, and that GMI's concepts were observed among referenced sources in AI summaries. While causal inference is not possible, the case illustrates that clear concept definitions and stable hosting can coincide with observable reference signals in generative search, even without high traffic. This finding contributes an empirically grounded instance to ongoing discussions of generative engine optimization.