Originally published at chudi.dev

AI search engines do not rank pages. They score entities, then quote the entity that best matches the query. For a sub-DR-20 brand, this is the good news. You cannot outspend enterprise SEO teams on backlinks. You can out-engineer them on entity coherence.

This post is the engineering playbook for that work. It covers the three schemas that anchor the graph, the sameAs surface where most brands leak trust, the knowsAbout field that tells engines what you are an authority on, and the measurement loop that tells you when it is working. The reference model throughout is chudi.dev (the Person brand) and citability.dev (the product brand): a deliberate two-node graph that synergizes without cannibalizing.

§1 — Why AI citations resolve to entities, not URLs

Classical SEO optimized for pages. Answer engines optimize for the entity behind the page. If two sites say the same thing and one site has a coherent Person + Organization + sameAs graph, the engine quotes the coherent site even when the other page ranks higher in traditional SERPs. Cite the entity, not the URL. That is the mental model shift this cluster pivots on.

§2 — The three anchor schemas

Every resilient entity graph has three anchor schemas. Person. Organization. SoftwareApplication. They cross-reference through creator, publisher, and about fields. Miss one anchor and the graph reads as a disconnected individual, a disconnected company, or a disconnected tool. Engines pick the version of your story they find easiest to summarize, which may not be the version you want quoted.

§3 — sameAs coherence across platforms

The sameAs array is the single most valuable field in the Person schema for sub-DR-20 brands. It is also the field most brands let drift. Six platforms. Six subtly different job titles. Six subtly different descriptions. Engines treat this as ambiguity and choose a canonical, usually the platform with the highest authority, not the one you care about.

§4 — The knowsAbout surface nobody uses

knowsAbout is the declarative authority claim. Use it to stake your territory. For chudi.dev, the stake is AI Visibility Engineering, Generative Engine Optimization, Entity Graph Architecture, and Sub-DR-20 SEO. Four claims. Four phrases an engine can match against a query.

§5 — The citation flow

When a query arrives, an AI engine runs a pipeline. Retrieve candidate entities. Score coherence. Gate. Cite. Optimizing for citation means shortening the distance between query and gate.

§6 — Measurement: how you know it is working

The answer engines that matter (Perplexity, ChatGPT with search, Google AI Overviews) do not expose rank. They expose citations. Measurement is a different instrument than Google Search Console. Track citation count per engine, quote length per citation, and coherence-check failures across your sameAs graph.

Bridge — from thinking to measurement

This post describes the thinking. citability.dev is the instrumentation. Run the citability scorer on any chudi.dev post to see what the engine actually extracts, which entity it resolves to, and where the coherence gates are failing.