Discover Labs: AI search makes the LLM the new website visitor

The conversation, hosted by Daniel Whitenack and Chris Benson on the Practical AI Podcast, centers on a shift the guests say is already underway: people research and consume information inside AI chat interfaces rather than on websites. Dunne frames this as a change in buyer behavior with a measurable symptom — companies noticing their organic traffic disappearing as the chatbot, in his words, chews up and spits back the information a site used to host.

Moore adds the technical layer behind that shift. He describes the prompt being tokenized and reasoned over, a retrieval engine generating query fan-outs, and a consensus-style algorithm similar to one DeepMind developed filtering candidates before a final answer is written. In his account, citations and mentions in that final output are only the last stage of a longer pipeline, and he points to model weights and the model's internal reasoning as earlier places where a brand's visibility is decided.

Much of the discussion turns on measurement. Moore argues the SEO field has not approached AI visibility with the statistical rigor the problem demands, and he gives the example of treating a stock-index move as alarming without checking it against a year of standard deviation. Whether Discover Labs' four-pillar framing and its push to bound uncertainty become standard practice — or whether agencies keep buying off-the-shelf tools and watching numbers rise and fall — is the open question the episode leaves for marketers to weigh.