The shift you're describing has a name that more developers should know — GEO: Generative Engine Optimization. It's the evolution of traditional SEO specifically for AI-powered search surfaces like ChatGPT, Perplexity, Google AI Overviews, and Gemini. The core difference is this: traditional SEO optimizes for ranking — getting your page to appear in a list. GEO optimizes for citation — getting your content included inside an AI-generated answer. Two completely different goals requiring different strategies. For GEO specifically, a few things that matter beyond the JSON-LD and llms.txt you mentioned:
Quotable sentences — AI systems prefer content that contains clear, standalone, citable statements rather than long flowing paragraphs Entity clarity — your page should unambiguously define who you are, what you do, and what problems you solve, so an LLM can map you to a knowledge node Third-party mentions — LLMs weight what others say about you (forums, Q&A sites, LinkedIn posts) more heavily than your own website copy
The underlying principle: traditional SEO asks "can Google find me?" — GEO asks "can an AI accurately represent me in an answer?" Those are very different optimization targets.