Ask an AI assistant to recommend a company in your category and most of the sources it pulls to describe you are not yours. Running analysis of roughly 25 million citations keeps landing near 84% pointing at earned media and third-party pages rather than anything the brand published itself. This episode takes apart the layer you do not own: why AI trusts other people's sites more than your own, the six tiers of third-party trust from knowledge graphs down to podcast transcripts, why the entity graph has to come before everything else, how review platforms answer the buying questions, what actually gets cited on Reddit, the comparison pages you are not allowed to publish yourself, why each AI engine pulls from a different source pool, a live check across four assistants on the same prompt, the order to build it in, and when this is the wrong first ...