The experiment
Over three days in early July, we ran a systematic audit of ChatGPT's real estate agent recommendations for Miami. The prompt was simple, and one a real relocating buyer might type: "Who are the best luxury real estate agents in Miami right now?" — and variants across Miami Beach, Coral Gables, Coconut Grove, and the broader South Florida market.
We ran each query 10 times across different sessions and accounts, tracked which agents and brokerages got named, and looked at the sources ChatGPT was citing. The results say a lot about the current state of AI search in real estate — and what agents can actually do about it.
Finding 1 · Big brokerages dominate
ChatGPT names the brands, not the agents
Across every Miami luxury real estate prompt, the same three or four brokerage brands appeared: Douglas Elliman, ONE Sotheby's, Compass, and Serhant. These weren't necessarily the top-producing agents — they were the brands with the deepest citation profile across the web.
Individual agent names surfaced only when we prompted specifically for "Coral Gables" or "Coconut Grove." Even then, the same handful of names came up — people who had media coverage, press appearances, or podcast interviews indexed by search engines.
The takeaway for individual agents: if you're not a media presence beyond your own website, ChatGPT doesn't know you exist. This is the single biggest gap most solo luxury agents have.
Finding 2 · The sources are predictable
Zillow, Realtor.com, and press coverage do the heavy lifting
When ChatGPT surfaced sources for its recommendations, three types of pages dominated: Zillow and Realtor.com profile pages, luxury real estate publications (Robb Report, Mansion Global, Miami Herald real estate section), and podcast appearances.
Notably absent: individual agent websites. If an agent had a strong personal site but hadn't appeared in third-party press, the AI couldn't build enough context to recommend them confidently. This is the fundamental gap between traditional SEO and AI Search Optimization — the citations agents optimize for on their own site aren't the citations that get them named.
Finding 3 · Instagram influence hasn't translated yet
The agents with big Instagram followings aren't getting named
This was the surprise. We tested against a list of Miami agents with strong Instagram followings — 50K+ followers, active posting, real market presence — and almost none of them appeared in ChatGPT's answers. Social influence isn't feeding AI answers directly, because AI engines don't yet crawl Instagram deeply for reputation signals.
The lesson: Instagram builds direct audience; AI Search Optimization builds discovery. They're not the same channel and they don't cross-pollinate automatically. The agents winning in one are often invisible in the other. The agents who'll win the next five years will do both intentionally.
Finding 4 · Location specificity matters a lot
"Best in Miami" gives one answer. "Best in Coral Gables" gives another.
The broader the query, the more the AI defaulted to household-name brokerages. As soon as we narrowed to submarkets, the answers got more specific — and often more accurate. Agents who dominate a defined neighborhood have a real shot at being named for that neighborhood, even if they'd never rank for "Miami" as a whole.
This is the single biggest strategic opening for boutique agents: go deep in one submarket rather than shallow across the metro. An agent who owns Miami Beach oceanfront condos has better odds of AI recommendation for that niche than an agent doing a bit of everything across the county.
"The agents who'll win AI search in real estate aren't the biggest producers. They're the ones with clearest positioning and the most third-party citation."
— Andrii Orfani, Co-Founder
What this means if you're a Miami agent right now
The audit surfaced a few concrete plays that would move the needle for anyone reading this:
- Get on the third-party sources ChatGPT already trusts. Complete and rich Zillow and Realtor.com profiles. Push for at least one press mention per quarter — even in local trade publications.
- Pick one submarket and own it in your content. "Miami real estate" is impossible to win. "Coral Gables historic homes" is winnable in 6-12 months of focused work.
- Publish content that directly answers the questions buyers actually ask AI. "What are the best neighborhoods in Coral Gables for young families?" gets typed into ChatGPT. Write the answer.
- Get on podcasts, especially local ones. Podcast transcripts feed AI models more than most agents realize.
- Layer Instagram + AI Search — don't pick one. The former builds relationships, the latter builds discovery. Both compound over time.
The bigger picture
AI Search Optimization for real estate is still an open field. Less than 5% of agents have any strategy at all, and even the ones who are optimizing tend to focus on their own sites while ignoring the third-party citation signals that actually feed AI answers. That's the arbitrage.
The Miami market we audited is competitive, but the field of agents actively working on AI visibility is measurable in single digits. In 12-18 months, that won't be true. Right now, being early is a real advantage.
If you want a similar audit for your name or your brokerage — we run these as free visibility audits. We'll tell you exactly which competitors ChatGPT is naming instead of you, and what it would take to fix.