AI SEARCH VISIBILITY
INTEL
Powered by AgentAir · agentairdirect.com
🔗 agentairdirect.comReal-time tracking of AEO · GEO · AI Mode · Answer Engine shifts
1B
AI Mode Monthly Users
2.5B
AI Overviews MAU
3x
Longer AI Queries vs Trad.
11%
Cross-Platform Citation Overlap
Product Architecture
THE AGENTAIR MODULE STACK
9-Module AI Search Visibility Infrastructure — Built to Find, Fix, and Monetize AI Invisibility
Perplexity-Ready AI Visibility Audit
Scans business websites specifically against Perplexity Sonar's citation signals — NAP coherence, FAQ schema, entity clarity, crawlability, freshness
Sonar Citation Score + Critical Fix Plan
Bulk AI Search Visibility Audit
Runs AI visibility audits across multiple businesses or locations simultaneously — built for agencies managing 10–50+ clients at once
Portfolio-Level AI Visibility Report
Entity Schema + FAQ Intelligence
Converts business data into machine-readable JSON-LD with LocalBusiness schema, sameAs entity stitching, FAQPage markup, and Organization schema
Copy-Paste JSON-LD Schema Code
Auditus: AI Explainer
Translates complex AI search visibility audit findings into plain-language explanations any business owner can understand — no tech background required
Plain-English Audit Breakdown
Archon: Compliance & Trust Shield
RAG-powered scoring for GDPR, ADA/WCAG, privacy policy, ToS, HTTPS, cookie notice, and business legitimacy — both a legal defense and a Sonar trust signal
Risk Assessment + Dev Checklist
Local Intent & Freshness Engine
Maps local intent gaps, Foursquare accuracy, listing consistency, and content freshness signals — ensures the business is found for geo-specific AI queries
GEO Score + Auto Local Schema
AuditVoice: Perplexity Sales Scripts
Generates Perplexity-specific sales scripts, cold call sequences, 3-step email drips, and 30-day follow-up language from live audit data using the "Workhorse" persona
Cold Call Script + 3-Step Email Sequence
Agency Intelligence Dashboard
Portfolio-level analytics for agencies managing 50+ clients — visibility score distribution, compliance trends, opportunity segments, exportable PDF reports
Portfolio Analytics + PDF Reports
Quote Calculator
Calculates the exact cost of AI visibility repairs based on specific audit failures — turns technical findings into a priced service proposal automatically
Automated Price Quote
Deep Protocol Analysis
PERPLEXITY AI CITATION PROTOCOL
60 Million Users. Zero Keyword Rankings. One Rule: Get Cited or Stay Invisible.
HOW PERPLEXITY'S SONAR ENGINE ACTUALLY WORKS
The 5-step retrieval pipeline that decides who gets cited
Step 1
QUERY EXPANSION
User prompt is rewritten into 3–5 parallel sub-queries by the Sonar model
Sonar InternalStep 2
LIVE RETRIEVAL
Each sub-query hits the live web. ~10 candidate URLs retrieved per sub-query in real time
100B URL IndexStep 3
RE-RANKING
Retrieved set ranked by: source authority, freshness, snippet quality, semantic relevance
Sonar RankingStep 4
PASSAGE SELECTION
Each paragraph evaluated independently as a 'semantic payload.' Top passages extracted
Vector EmbeddingStep 5
CITATION SYNTHESIS
Top 3–4 URLs become visible citations. Answer synthesizes facts inline with footnote links
Answer Output🎯 THE FUNNEL: Retrieve 10 URLs → Cite 4. You don't need to be #1. You need to be in the top 4 of the retrieval set — a fundamentally different game from Google ranking.
THE 12 PERPLEXITY CITATION SIGNALS
Every Signal Sonar Uses to Decide Who Gets Cited — not SEO guesses, but protocol-level evaluation criteria
PERPLEXITY vs GOOGLE: WHY BUSINESSES ARE MISSING 60M CUSTOMERS
Side-by-side comparison of two fundamentally different search architectures
| Dimension | Google Search | Perplexity AI Sonar |
|---|
THE AGENTAIR ↔ PERPLEXITY ALIGNMENT MAP
AgentAir IS What Perplexity Needs From Every Business
Every Sonar citation signal has a direct AgentAir module that audits, fixes, and optimizes for it.
| Perplexity Signal | What Perplexity Needs | AgentAir Module | What AgentAir Delivers |
|---|
Golden River Connect™ ↔ Sonar
Golden River Connect™ is AgentAir's answer to Perplexity's Sonar retrieval pipeline. While Sonar reads the web and decides who to trust, Golden River Connect™ builds the trust infrastructure before Sonar arrives — making your business machine-readable, entity-verified, citation-ready across every platform Perplexity pulls from.
THE BOTTOM LINE: 3 THINGS PERPLEXITY NEEDS FROM EVERY BUSINESS
BE FINDABLE
Perplexity runs a live search. If your business isn't indexed, listed on review platforms, or mentioned in any authoritative source — it doesn't exist to Sonar. AgentAir's AEO Audit + GEO Mapping Engine fixes this.
BE READABLE
Sonar evaluates each paragraph as a standalone semantic payload. If your business data isn't structured in machine-readable JSON-LD with consistent NAP — Sonar can't confidently cite you. AgentAir's Entity Schema Engine fixes this.
BE TRUSTWORTHY
Sonar has a legitimacy verification layer. No HTTPS, no privacy policy, no terms of service, no consistent contact info = low confidence source = no citation. AgentAir's Compliance Shield fixes this.
Deep Protocol Analysis
CHATGPT SEARCH CITATION PROTOCOL
900M Weekly Users. 87% of All AI Referral Traffic. Two Citation Gates — Both Must Open.
THE TWO-LAYER CITATION SYSTEM: HOW CHATGPT ACTUALLY WORKS
ChatGPT does NOT rank pages. It runs a two-layer citation system most businesses never learn about — and failing either layer means zero visibility, no matter how good your Google SEO is.
Layer 1
Parametric Knowledge Layer
ChatGPT's base model was trained on a massive web corpus. If your brand appeared frequently in high-quality sources during training — Wikipedia, major publications, Reddit threads, industry reports — the model knows you exist. This is your AI brand fingerprint. You can't change past training data, but you can build the web presence that gets included in future training runs.
ALWAYS ON — NO SEARCH TRIGGEREDKey signal: Entity recognition, Wikipedia presence, brand mentions in authoritative sources
Layer 2
Retrieval-Augmented Generation (RAG) Layer
When a query needs current data, ChatGPT switches to Browse Mode. It queries Bing's index in real time, retrieves candidate pages, evaluates each one, and selects sources to cite. This is where your structured data, technical setup, and content architecture determine whether you get cited. Only 15% of retrieved pages ever make it to citation — the other 85% are evaluated and discarded.
TRIGGERS ON DISCOVERY + COMPARISON QUERIESKey signal: Bing index, OAI-SearchBot access, structured data, content freshness
🎯 THE HARD TRUTH: ChatGPT cites only 15% of the pages it retrieves (AirOps, 548,534 pages, 15,000 prompts, March 2026). There are TWO optimization problems: (1) Discoverability — getting into the retrieval pool, and (2) Selectability — surviving the cut. Most businesses only know about the first problem.
THE 6-STEP CHATGPT SEARCH PIPELINE
The retrieval pipeline that decides who gets cited — and who gets discarded
Step 1
SEARCH TRIGGER DECISION
ChatGPT decides: does this prompt need web search? Discovery queries trigger search 73% of the time. Informational 'what is' queries trigger only 10%.
PARAMETRIC vs RAG DECISIONStep 2
BING INDEX QUERY
OAI-SearchBot queries Bing's index. 73% overlap between ChatGPT citations and Bing top results. No Bing index = zero chance of citation.
BING IS THE GATEKEEPERStep 3
SUB-QUERY EXPANSION
Query rewritten into multiple sub-queries. Discovery and comparison queries trigger the highest sub-query expansion rates.
MULTI-VECTOR RETRIEVALStep 4
PAGE RETRIEVAL + RENDERING
Candidate pages retrieved and rendered server-side. JavaScript-dependent pages are invisible to GPTBot and OAI-SearchBot. SSR is non-negotiable.
SSR REQUIREMENTStep 5
CONTENT EVALUATION (The 85% Discard)
Each page evaluated for: relevance, source authority, content clarity, extractability, and freshness. 85% are discarded. Only 15% survive to citation.
THE SELECTABILITY GATEStep 6
CITATION + SYNTHESIS
Top 2-4 URLs become footer citations. Homepage referrals surged 354.7% after ChatGPT added clickable brand links (May 7, 2026).
ANSWER OUTPUT🎯 THE QUERY TYPE FUNNEL: Discovery queries ('best X for Y') trigger web search 73% of the time. Comparison queries ('X vs Y') trigger 65%. Generic informational queries ('what is X') trigger only 10%. If your content only targets 'what is' and 'how to' keywords — you're competing in the 10% zone. The 73% zone belongs to discovery and comparison content.
OPENAI'S THREE CRAWLERS — EACH DOES A DIFFERENT JOB
Blocking the wrong one silences you. Knowing the difference is a competitive advantage.
GPTBot
User-agent: GPTBot/1.1
Crawls content for training OpenAI's foundation models. Block this if you don't want your content used in model training. Blocking GPTBot does NOT affect your ChatGPT Search citations — these are completely separate systems.
Disallow: /
OAI-SearchBot
User-agent: OAI-SearchBot/1.0
This is the bot that indexes your content for ChatGPT Search citations. If you block OAI-SearchBot, you have zero ChatGPT Search citation eligibility — permanently. Must be allowed in robots.txt. IndexNow integration gets pages indexed in 12-48 hours vs 14-30 days without it.
Allow: /
ChatGPT-User
User-agent: ChatGPT-User/1.0
Triggered when a human user asks ChatGPT to visit a specific URL. Not automated crawling — purely user-initiated. Blocking this prevents ChatGPT from fetching your pages during live user conversations.
Allow: /
# Recommended: Get cited, stay out of training
User-agent: GPTBot
Disallow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
THE 12 CHATGPT CITATION SIGNALS
Sourced from AirOps (548,534 pages), SE Ranking, Authoritas, Fortis Media, Lee 2026 (19,556 queries). These are the exact protocol terms.
CHATGPT vs GOOGLE vs PERPLEXITY: WHY YOUR GOOGLE STRATEGY IS MISSING 87% OF AI REFERRAL TRAFFIC
| Dimension | Google AI Overviews | Perplexity Sonar | ChatGPT Search |
|---|
THE AGENTAIR ↔ CHATGPT ALIGNMENT MAP
Golden River Connect™ Is AgentAir's Answer to ChatGPT's Entity Graph
Every ChatGPT citation signal has a direct AgentAir module that audits, fixes, and optimizes for it.
| ChatGPT Signal | What ChatGPT Needs | AgentAir Module | What AgentAir Delivers |
|---|
Golden River Connect™ — The Entity Graph Builder
While ChatGPT's entity graph evaluates your brand across Wikidata, Wikipedia, Crunchbase, Reddit, and industry publications, Golden River Connect™ maps how AI reads, interprets, and verifies your business entity across every platform — building the trust infrastructure ChatGPT needs before it will confidently cite you.
THE BOTTOM LINE: 3 THINGS CHATGPT NEEDS FROM EVERY BUSINESS
BE IN BING
ChatGPT's citation pipeline starts at Bing's index. Google rankings are irrelevant. If OAI-SearchBot is blocked or your site isn't in Bing Webmaster Tools, you don't exist to ChatGPT Search — for any query, forever. AgentAir's AEO Audit Engine fixes this in the first scan.
BE AN ENTITY
ChatGPT doesn't cite websites. It cites recognized entities. Your brand must exist in Wikidata, Wikipedia, Crunchbase, LinkedIn, and G2 — with sameAs schema stitching them all together. r=0.76 correlation between entity density and citation. AgentAir's Entity Schema Intelligence Engine builds this infrastructure.
BE SELECTABLE
85% of retrieved pages are discarded before citation. The selectability gate is won by: FAQ schema (40% lift), atomic paragraphs with 40-60 word front-loaded answers, content freshness within 30 days, and discovery query targeting. AgentAir's AEO Audit Engine scores every one of these signals.
Intelligence Log
MILESTONE TIMELINE: MAY 2026 → NOW
Google Drops 5-Feature Citation Overhaul for AI Mode
Biggest citation-surface update since AI Overviews launched. Inline links, hover previews, Subscribed badge, Community Perspectives quotes, Explore Further module.
Google VP Hema Budaraju announced 5 simultaneous changes to how AI Mode and AI Overviews link out. Inline links now appear next to relevant generated text. Hover previews show destination site name and page title on desktop. A "Subscribed" badge highlights links from publications users already pay for. Community Perspectives quotes Reddit, forums, and expert blogs. Explore Further module points to deeper adjacent reads. Context: AI Overview presence already correlating with ~58% lower CTR for top-ranking pages per Ahrefs data.
AGENT INFRA PARALLEL: Citation routing in AI answers now mirrors agent tool-call citation — sources are ranked, filtered, and surfaced by confidence score, not just index ranking.
Google Officially Declares: AEO + GEO = "Still SEO"
Google publishes new documentation page "Optimizing your website for generative AI features on Google Search" — directly addresses AEO/GEO debate and kills several popular myths.
Google's guide states: "From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO." AI features are rooted in core Search ranking and quality systems and rely on RAG (retrieval-augmented generation) and query fan-out. Google explicitly names tactics that are NOT needed: llms.txt, content chunking for AI, AI-specific rewriting, special schema for generative features, inauthentic mentions. Gary Illyes and Cherry Prommawin at Search Central Live had already said GEO/AEO don't need separate frameworks — this guide makes it official documentation.
AGENT INFRA PARALLEL: Google's RAG + query fan-out architecture is essentially an agent retrieval pipeline — the same infrastructure that powers autonomous agents (plan → retrieve → synthesize → respond) is now powering standard search answers.
Google I/O: "Google Search IS AI Search" — 1 Billion Users, Agents, New Box
Google's biggest Search announcement in 25 years. AI Mode hits 1B MAU. Gemini 3.5 Flash becomes default. Search box fully redesigned. Information Agents previewed for summer 2026.
Key announcements: (1) AI Mode surpassed 1 billion monthly users in its first year, with query volume doubling every quarter. (2) Gemini 3.5 Flash is now the default model powering AI Mode globally. (3) The search box received its biggest redesign in 25+ years — renamed context shifts toward "Ask Google," dynamically expands, supports text/image/file/video/tab inputs, AI-powered suggestion beyond autocomplete. (4) Information Agents: always-on background AI agents that monitor the web continuously on users' behalf without them typing a query — coming summer 2026 for AI Pro+ subscribers. (5) Seamless flow between AI Overviews and AI Mode across desktop and mobile. Google VP Liz Reid: "Google search is AI search."
AGENT INFRA PARALLEL: Information Agents ARE agent infrastructure — they run background tasks, monitor web state changes, and surface findings proactively. This is the same architecture as autonomous task agents: persistent context, event triggers, retrieval, action.
Google May 2026 Core Update Begins Rolling Out
Second core update of 2026. Rolls out mid-I/O week, creating overlapping signal noise for SEOs. Targets sites over-optimizing for AI citations.
Google announced via Search Status Dashboard. Key context: This is the second core update of 2026 (follows March 2026 core update, March 2026 spam update, February 2026 Discover update). Timeline overlap with I/O announcements makes it hard to isolate what caused ranking shifts. SEO consultant Harpreet Singh Chatha predicted this targets AI citation over-optimization. Marie Haynes connected it to Gemini 3.5 Flash powering search features. Google advice: create helpful content for people, not engines.
AGENT INFRA PARALLEL: Core updates increasingly reflect model behavior updates, not just link-graph recalculations. As Gemini models are updated, the retrieval quality of AI Overviews changes — same dynamic as updating an agent's base model.
Major Ranking Volatility Wave — Core Update Second Spike
Significant secondary volatility wave from May core update. Sites seeing substantial gains and losses as update propagates through Google's index.
Second major spike in ranking volatility from the May 2026 core update. First wave hit weekend of May 23-24; second major wave hit May 30. Semrush volatility charts show back-to-back spikes. Some recovery possible between core updates, but biggest changes typically follow another core update. No specific recovery actions available per Google guidance.
AGENT INFRA PARALLEL: Model rollouts in agent infrastructure show similar phased propagation — updates don't land uniformly, causing inconsistent behavior across regions and query types before stabilizing.
May 2026 Core Update Confirmed Complete
Google confirms rollout complete after 12 days. Final volatility spike observed in final 24 hours before completion.
Google updated Search Status Dashboard confirming completion. The update ran May 21 – June 2, approximately 12 days. Even more volatility observed in final 24 hours before completion — unusual for typical core update wind-down. Healthcare ads testing begins in AI Mode. AI Mode and AI Overviews testing sliding carousel for citation link cards. Google advice remains: write helpful, reliable, people-first content.
AGENT INFRA PARALLEL: Agent infrastructure stabilization post-deployment mirrors this — final inconsistencies resolve as caches clear, model routing stabilizes, and retrieval indexes sync globally.
Architecture Analysis
THE PARALLEL STACK: AEO Infrastructure ↔ AgentAir Infrastructure
| Concept | AEO / AI Search Visibility | Agent Infrastructure Parallel | AgentAir Module |
|---|
Copied to clipboard!
Platform Intelligence
WHERE YOUR CONTENT GETS CITED
AI Search Platform Citation Profiles (2026)
ChatGPT Search
OpenAI · Bing Index
Citation Bias
Optimization Tip
Schema markup, content freshness, branded authority sites
Perplexity
Brave Search + Proprietary
Citation Bias
Optimization Tip
Citation density, content specificity, factual accuracy
Google AI Mode
Full Google Index · Gemini
Citation Bias
Optimization Tip
E-E-A-T, structured answers, entity clarity, schema
📡 Signal Feed
DAILY INTEL FEED
Auto-updated every morning with the latest AI search visibility developments
- ✅May 2026 Core Update confirmed complete (12-day rollout)
- 🔬Google testing healthcare ads inside AI Mode — first monetization signal for AI Mode
- 🎠AI Mode + AI Overviews testing sliding carousel for citation link cards
- 📊AI Overviews CTR impact: -38% on triggered queries (Pew/SEL data)
- 🤖Information Agents for AI Pro+ still on track for summer 2026 release
- ▸Google I/O: AI Mode query volume doubling every quarter
- ▸May 2026 Core Update launched, overlapping with I/O week
- ▸Google Gemini 3.5 Flash now default model in AI Mode globally
- ▸New Search box rollout begins worldwide
- ▸Google I/O 2026 Day 1: "Google search IS AI search"
- ▸1B monthly AI Mode users confirmed
- ▸Information Agents announced for summer 2026
- ▸Biggest Search box redesign in 25 years