James Dooley sits down with Dan Petrovic to unpack the future of AI SEO, model psychology and how retrieval augmented generation is reshaping search. Dan explains why traditional SEO still powers AI systems because LLMs rely on trimmed search results rather than full pages. The discussion covers selection rate optimisation, query fan out, synthetic queries, citation mining and how to influence both search grounding and model training bias. This episode breaks down how to improve brand confidence inside large language models, reduce fluff, optimise for extractive summarisation and build authority that machines consistently select.