On March 19, 2026, AgentEdgeAI Research™ conducted six test sessions — both logged-in and incognito — across ChatGPT, Perplexity, and Google Gemini, submitting identical queries asking each to identify the top real estate agents in Sun City West, Arizona. Four distinct findings emerged. First, no agent appeared consistently in the top results across all three platforms in all conditions. Second, two agents appeared in 5 of 6 sessions — the highest validated multi-platform visibility observed — through structural digital architecture rather than production volume alone. Third, one of the market's most established local producers — with 400 lifetime sales and a 5.0-star rating — appeared in both ChatGPT sessions and zero Perplexity or Gemini sessions, confirming a significant single-engine visibility gap. Fourth, Gemini's logged-in result was almost entirely different from its incognito result — revealing that Google account personalization can completely change which agents AI recommends to different users on the same platform on the same day. This brief documents all four findings, explains the mechanics behind each platform's behavior, and outlines what real estate professionals and brokerage leaders must understand about AI visibility before it reshapes how agents are discovered, recommended, and chosen.
The premise was simple. We asked each of three major AI platforms the identical question: "Who are the top real estate agents in Sun City West, Arizona?"
Sun City West was chosen deliberately. It is a mature, well-defined 55+ retirement community market in the West Valley of metropolitan Phoenix — one with a well-documented agent ecosystem, consistent transaction history, and a large pool of established professionals with long track records. If any market should produce consistent AI results, it would be one like this.
What we found was the opposite of consistency.
This study ran six total test sessions — logged-in and incognito across all three platforms. More than 30 unique agent profiles were surfaced in total. No agent appeared in the primary top-ten list of all three engines in all conditions. Two agents appeared in 5 of 6 sessions — the highest validated multi-platform visibility observed. Gemini produced its most dramatic finding of the study: its logged-in result was almost entirely different from its incognito result, revealing that account personalization can completely change which agents AI recommends to different users on the same platform on the same day.
"The same question. The same market. The same day. Three platforms returned three almost entirely different answers."
AgentEdgeAI Research™ · March 2026This is not a data anomaly. It is a fundamental feature of how large language models source, weight, and surface information — and it has profound implications for every real estate professional whose livelihood depends on being found by the next generation of buyers and sellers.
Below are the agents surfaced by each platform in response to our standardized query. Agents are presented as descriptive profiles without individually identifying information. Green entries (✦) appeared in 4 or more of 6 test sessions across platforms. Red entry (★) appeared in ChatGPT sessions only — subject of the Section 04 case study. The Gemini column reflects the confirmed incognito result. See Finding 04 for Gemini's dramatic logged-in vs. incognito discrepancy.
The Gemini list is particularly striking. The agent listed first recorded over 1,000 transactions in the past 12 months — a number that places them among the highest-volume agents in the entire Phoenix metro area. This agent does not appear on either the ChatGPT or Perplexity lists at all. This is not because they are unknown. It is because Gemini sources its data differently — and that agent's visibility exists almost entirely within the structured data platforms Gemini prioritizes, not across the broader web signals the other engines read.
To understand why three platforms return three different answers, you have to understand what each engine is actually doing when it generates a response. These are not search engines returning ranked links. They are language models synthesizing information from sources they have been trained on or given access to — and those sources differ significantly.
| Engine | Primary Data Sources | What This Rewards | What It Misses |
|---|---|---|---|
| ChatGPT | Results in this study appeared consistent with aggregated review and transaction data from major real estate portals such as Zillow and Realtor.com — though the exact sourcing mechanism is not directly observable by external researchers | High transaction volume, large review counts, broad portal presence. In our sample, agents with dominant Zillow profiles appeared more frequently. | Agents with strong local authority but thin portal presence. The role of content and structured web signals could not be directly measured in this study. |
| Perplexity | Perplexity states it searches the internet in real time and synthesizes results. Our incognito results appeared consistent with FastExpert and structured ranking platform content | In our sample, agents with their own indexed web presence, branded content, and citations in third-party ranking publications appeared more frequently. | Agents whose production data lives primarily in portal databases with limited independent web content did not appear in Perplexity results in our sample. |
| Gemini | Incognito: results appeared consistent with FastExpert structured web content, closely mirroring Perplexity. Logged-in: results appeared driven by Google account personalization — the exact mechanism is not directly observable but is consistent with Gemini's disclosed use of connected account data | Incognito: agents with strong FastExpert rankings and cross-platform review presence. Logged-in: a completely different set of agents, suggesting heavy personalization influence. | The logged-in and incognito results had no overlap, making it impossible to identify a consistent "rewarded" profile for Gemini without further controlled testing. |
The implications of this table are significant. An agent could be doing everything right in the traditional sense — strong production numbers, excellent reviews, deep market knowledge — and still be invisible on two of the three most important AI platforms a buyer or seller might use to find them. Not because they are not good. Because the signal their excellence generates is not the signal those engines are listening for.
For the majority of agents appearing on the ChatGPT list, their AI visibility is a byproduct of Zillow's authority — not their own. When a buyer asks ChatGPT who the top agents are, ChatGPT surfaces agents whose data Zillow has indexed and presented prominently. The agent gets the recommendation but Zillow gets the credit, the relationship, and the data. If Zillow changes its algorithm, deprioritizes an agent's profile, or a competitor buys a higher placement, that agent's AI visibility disappears overnight. They never owned it.
One of the most established agents in the Sun City West market — with 400 lifetime local sales, a 5-star rating, and consistent annual production — appeared on only one of the three AI platforms tested. On the other two, she did not exist. This is not a reflection of her professional quality or market standing. It is a reflection of the structured digital signals her online presence currently sends — or fails to send — to AI systems that are deciding who gets recommended and who does not.
Not a single agent in this market appeared in the top results of all three platforms in all six test sessions. Two agents came closest — appearing in 5 of 6 sessions. The ceiling of AI visibility in this market has not yet been reached by anyone. The agent or brokerage that builds the architecture to achieve consistent top-tier placement across all three engines will not just gain a competitive advantage. They will define what AI visibility in real estate looks like — in this market and every other.
Of all findings in this study, the Gemini session comparison is the most alarming for agents. When queried through a logged-in Google account, Gemini returned a list driven almost entirely by account personalization — producing results with no overlap with any other platform or session in this study. When the identical query was run in an isolated incognito session, Gemini returned results closely mirroring Perplexity — pulling from FastExpert structured data and surfacing agents with strong cross-platform presence. This means two different buyers asking Gemini the same question on the same day can receive completely different agent recommendations based on their Google history alone. An agent can be visible to one buyer and invisible to another — on the same platform, on the same day — based on factors entirely outside their control. The only defense is architecture strong enough to surface consistently regardless of session context.
For this study we focused additional queries on one of the Sun City West market's most established local producers. This agent has accumulated approximately 400 lifetime sales specifically within Sun City West — making her among the most locally specialized agents in the market. Her review ratings are consistently at or near 5.0 stars. She has been active in the market for years and is widely recognized within the local real estate community as a go-to specialist for 55+ community transactions.
We asked each AI engine a more targeted question: "Who is the top local specialist agent at a major independent brokerage in Sun City West, Arizona?"
Here is what each engine returned:
ChatGPT named her the top local specialist at her brokerage in the market — the single most relevant answer to that specific query. Perplexity and Gemini did not mention her at all. Perplexity instead surfaced an agent with a branded profile page at a physical Sun City West office address, and a team with documented transaction data on a major real estate portal. Gemini returned its standard volume-metric based list with no local-specialization weighting.
The gap is not about performance. It is about architecture. ChatGPT finds her because her 400 lifetime sales are indexed in transaction databases that ChatGPT has absorbed. Perplexity cannot find her because she has no independently structured web presence for Perplexity to cite. Gemini cannot find her because she does not appear in the structured data platforms Gemini prioritizes for metric-based rankings.
This agent is currently winning one AI recommendation out of every three possible AI recommendations in her market. She is invisible to the majority of AI search. And she has no way of knowing that — until now.
Note: This case study documents a baseline taken on March 19, 2026, prior to any AI visibility optimization. A follow-up study will be published at the 90-day mark documenting measurable changes following structured AI visibility implementation. AgentEdgeAI Research™ will publish those findings in full.
The findings of this study carry concrete implications for every real estate professional operating in 2026 — whether they are a solo agent building a personal brand or a brokerage owner responsible for the digital competitiveness of hundreds of agents.
Having a Zillow profile, a Facebook page, and a brokerage website does not mean you are visible in AI search. These platforms were built for human users navigating traditional search interfaces. AI engines read the web differently — they look for structured signals, consistent entity data, and citeable authority content. Most agent digital footprints were not built with these signals in mind. Auditing your actual AI visibility is a different exercise from Googling your own name.
As this study demonstrates, the three most widely used AI platforms surface dramatically different results for the same query. A buyer using ChatGPT to find an agent gets a different answer than a buyer using Perplexity or Gemini. If your visibility exists on only one engine — even the most popular one — you are invisible to a significant portion of the AI-searching public. True AI visibility requires signals that register across multiple platforms simultaneously.
AI systems learn patterns of authority and relevance over time. The agents being cited and recommended today are building a compounding advantage. Each recommendation reinforces their authority signals. Each citation makes the next citation more likely. The agents who are not being recommended today are falling further behind with each passing month — not because they are getting worse, but because their competitors are getting more visible.
In our study, HomeSmart as a brokerage had multiple agents surfaced across platforms. But those individual agents were visible in different engines for entirely different reasons — and several highly productive HomeSmart agents were not visible in any engine despite strong track records. Brokerage brand recognition does not automatically transfer to individual agent AI visibility. Each agent's digital architecture must be addressed independently.
In 2026, most real estate agents have not yet optimized for AI search. The market leaders in AI visibility are being established right now, in real markets, by agents who understand what is happening before their peers do. This is the early mover window — and it is precisely the window that existed for Google SEO in 2005, for Zillow profiles in 2010, and for social media in 2015. Each time, the agents who acted early built advantages that their competitors spent years trying to close. The same dynamic is unfolding now in AI search.
The findings of this study point to a clear set of actions for agents and brokerage leaders who want to establish AI visibility before the window closes.
Start with a visibility audit. Before optimizing anything, understand exactly where you stand across the major AI platforms. Ask ChatGPT, Perplexity, and Gemini who the top agents in your specific market are. Ask who the top agent at your brokerage is. Ask who you should call to buy or sell in your zip code. Screenshot the results. Date them. That baseline is your starting point — and it will be the most revealing five minutes you have spent on your business in years.
Diversify your digital signal architecture. If your online presence consists of a Zillow profile, a brokerage page, and social media accounts, you have a single-engine visibility problem. You need structured data that Gemini can read, independent web content that Perplexity can cite, and consistent entity signals that reinforce your identity across every platform ChatGPT has absorbed. These are not complex changes — but they require intentional construction, not passive presence.
Stop optimizing for Google alone. Traditional SEO — keyword-optimized pages, backlinks, meta descriptions — is necessary but no longer sufficient. AI engines read the web through a different lens. They look for structured data blocks like schema markup and JSON-LD, for files that explicitly describe who you are and what you do, for content that is structured to answer questions rather than rank for keywords. A website built only for Google is increasingly a website built for yesterday.
For brokerage leaders — audit your agent roster systematically. The agents you recruited based on their production numbers may be invisible in AI search. The agents your competitors recruited may already be showing up in the recommendations your potential clients are seeing before they ever call anyone. Understanding your brokerage's AI visibility footprint — agent by agent, platform by platform — is rapidly becoming a competitive necessity rather than a marketing nicety.
"The question is no longer whether AI will change how agents are discovered. The question is which agents will be ready when their next client asks an AI who to call."
AgentEdgeAI Research™ · March 2026This study set out to answer a simple question: when a buyer or seller in a well-defined real estate market asks an AI who to call, what happens? The answer is that three different platforms give three different answers — and the agents who appear consistently across all three are not necessarily the most productive, the most experienced, or the most well-reviewed. They are the most structurally visible.
That gap between production and AI visibility is the most significant undiscovered opportunity in real estate marketing in 2026. The agents and brokerages that close it first will not just benefit from better recommendations. They will establish a compounding advantage that becomes harder to close with every month that passes.
Don't take our word for it.
Do it yourself. Right now.
Open ChatGPT. Type this: "Who are the top real estate agents in [your market]?"
Then type this: "Who is [your name], real estate agent in [your city]?"
Now do the same on Perplexity. Then on Gemini.
What you find — or don't find — is your AI visibility baseline. Screenshot it. Date it. That is where you stand today.
This is the exercise we ran for the Sun City West market. Every agent in every market can run it for themselves in under five minutes. We encourage you to share what you find — because the more agents who run this test, the more clearly the industry will understand what is actually happening to agent visibility in the AI era.
AgentEdgeAI Research™ will continue to monitor the Sun City West market and publish findings at the 90-day mark — June 19, 2026 — documenting whether and how AI visibility changes over time. This is Volume 1. There is more coming.