From zero PR to multi-LLM recognition.

What was tested

The question was simple: could a clear entity, structured proof, and answer-shaped pages make an AI system describe and recommend a person accurately starting from effectively zero third-party PR? The test tracked how multiple AI systems answered brand, category, and buyer-intent prompts over time.

Platforms and prompts

The prompt set covered entity prompts ("who is X"), category prompts, and buyer-intent prompts across the AI surfaces buyers actually use. Each answer was captured as a screenshot, with attention to accuracy, whether the entity was named, and which sources were cited.

The baseline

Before the work, the baseline was the honest starting point: scattered signals, proof living in private channels, and little for any model to retrieve. That baseline is what makes any later movement meaningful rather than anecdotal.

What signals were built

The build followed the AI Authority Protocol: a clean entity page, structured offer pages, a public proof archive, FAQ and definition clusters, internal links, and schema. Nothing here is a trick it is structure a machine can parse and corroborate.

Limitations and what this is not

No one controls what an AI system outputs, and no result here is guaranteed or repeatable on demand. AI answers change, vary by session, and depend on factors outside any one site. This case study documents signals built and changes observed it does not promise citations or rankings.

What this proves

It proves a narrower, more useful point: when an entity is clear and its proof is public and structured, AI systems have something accurate to retrieve and cite. Structure does not guarantee outcomes, but its absence almost guarantees invisibility. That is the difference the work targets.

How this connects to the system

This is the AI Authority Protocol applied to a real entity, measured with the method behind the AI Visibility Audit. The self-implementation version lives in the AI Authority System.