The ARDI™ Research Center.
GOSH AI's central body of research on AI discovery behavior and large language model citation patterns, based on real prompts, real outputs, and real-world consequences across the leading AI models. The research foundation behind the ARDI™ Intelligence Platform.
Observational, not predictive. Built to surface patterns so organizations can adapt with clarity.
New to AI discovery intelligence?
The ARDI™ Knowledge Base is the foundational reference: what it is, how AI models make recommendation decisions, why traditional optimization falls short, and what the shift from search results to AI verdicts means for your brand. Whether you are a CMO hearing about AI visibility for the first time or a strategist building the internal case, begin here.
Explore the Knowledge Base →Topics covered
- How AI models decide what to recommend
- Volume, Sentiment & Positioning explained
- The Think, Discovery, Decision journey
- Why brands disappear from AI responses
- What the ARDI™ Authority Score measures
- Common misconceptions about AI optimization
Five active lines of inquiry.
Each track observes a different dimension of how AI discovers, cites, and recommends brands. Every one maps back to the Four Authority States and Two-Path Presence.
ARDI™ Discovery Studies
"When a buyer asks AI about your category, which Authority State are you in?"
Field research documenting which brands AI cites, recommends, or excludes in AI-generated answers. Every observation maps to the Four Authority States and Two-Path Presence.
Enter the research →AI Model Behavior Observations
"How does AI decide what to say about your brand?"
How LLMs interpret prompts, resolve entity ambiguity, weigh source authority, and construct brand narratives, the mechanics behind why AI says what it says.
Enter the research →Real-World AI Failures
"What is AI getting wrong about you, and does it matter?"
Case studies where AI responses are inaccurate, misleading, or contextually wrong, and what they reveal about model limits, brand risk, and the cost of leaving AI perception unmanaged.
Enter the research →Voice Search & Conversational AI
"When someone asks out loud, are you the answer?"
How voice-prompted AI interprets brand queries differently from text, and the challenge of being the single spoken answer with no visual alternatives.
Enter the research →
What Is ARDI™?
"What is this new discipline, and why does it matter now?"
A reference overview defining AI Recommendation and Discovery Intelligence, how it differs from search optimization, and how the Four Authority States and Two-Path Presence operate. The conceptual foundation for everything here.
Read the overview →Every observation becomes a paper.
The full body of published ARDI™ research lives in the Intelligence Index: the State of AI Visibility volumes, market structure reports, and methodology briefs. Or see exactly where your brand stands with the free ARDI™ Snapshot.
Enter the Intelligence Index →How the research is done.
Nothing here is a hunch. Every observation is recorded, controlled, and reproducible.
Documented standards & controls
All research in the ARDI™ Research Center follows a documented methodology covering prompt controls, model selection, output recording, and interpretation limits. Every finding is traceable to specific prompts, models, and dates.
View the methodology →