How AI assistants decide which real estate agent to recommend

A growing number of buyers and sellers no longer start with Google. They open ChatGPT, Claude, Gemini, or Perplexity and type something like "who's a good real estate agent in Naperville for a first-time buyer?" And they get a name.

If your name isn't the one that comes back, you never find out. There's no ranking page to check, no analytics dashboard, no missed-call log. The recommendation happens in a conversation you're not part of.

So it's worth understanding what actually drives that answer.

The answer isn't a ranking, it's a reconstruction

Search engines return a list someone can audit. AI assistants do something different: they assemble a plausible answer from everything the model has read about your market. That means the "recommendation" is a reconstruction of your reputation as it appears across the open web: your brokerage page, local news, review sites, community forums, YouTube, press mentions, and how consistently your name shows up next to the things you actually do.

Two consequences follow from that.

First, consistency beats volume. An agent mentioned five times in ways that all say the same thing (first-time buyers, north-side condos, patient with nervous clients) is easier for a model to recommend confidently than an agent mentioned twenty times with no through-line. The model is pattern-matching a specialty, not counting backlinks.

Second, the answer moves. Ask the same question twice and you'll often get two different names. Ask it on Gemini instead of ChatGPT and you'll get different names again. One conversation is an anecdote, not a signal.

General visibility and specialty visibility are different problems

When we run visibility checks, we look at two separate questions, because agents are usually strong on one and invisible on the other:

• General visibility: does AI name you for a broad ask like "best agent in [your area]"? This is dominated by the largest, most-cited names in a market.
• Specialty visibility: does AI name you for the clients you actually want, like "an agent who's good with first-time buyers in [neighborhood]"? This rewards a clear, repeated niche far more than raw size.

Most agents assume they need to win the broad ask. In practice the specialty ask is both more winnable and closer to the client you want, because the person typing it is describing themselves.

What this means if you're building, not fixing

If AI barely knows you yet, that's not a verdict. It's a baseline, and an early one. Newer agents and agents with a small web presence almost always start with low AI visibility. The advantage of finding out now is that the signals AI reads are the same signals you can shape deliberately: a consistent description of who you help, in the places a model actually reads, saying the same thing every time.

That's the whole game. Not more noise: a clearer, more repeated story about the exact client you want to be recommended for.

---

Pitched runs live visibility checks across ChatGPT, Claude, Gemini, and Perplexity so agents can see where they appear, who appears instead, and what to improve. This post is part of an ongoing series on how AI search is reshaping how clients find agents.