How to Check If AI Recommends You (And What to Do While You Wait)

You can find out in the next five minutes whether AI is quietly sending business to your competitors instead of you. Most agents never check. They assume that if they've got good reviews and a decent website, the robots have it handled. Then a buyer says "I asked ChatGPT for the best agent in town and your name didn't come up" — and suddenly it's very real.

Here's how to check yourself, for free, today. And here's the honest part nobody tells you: the DIY check is a great start, but it's a snapshot, not a diagnosis.

The free check anyone can run right now

Open a private or incognito browser window. This matters more than anything else you'll do. If you ask an AI assistant while you're logged in, it already knows you — your searches, your location, your business — and it'll flatter you. You'll walk away thinking you're everywhere, when a stranger looking for an agent sees someone else entirely.

Incognito strips that out. Now you see what a customer sees.

Then ask the way a customer actually talks — not "am I ranked," but the messy, specific question a real person types:

• "Who's a good real estate agent in [your town]?"
• "I need to sell a house I couldn't sell on my own in [city] — who should I use?"
• "Best realtor for first-time buyers in [neighborhood]."

Run it on more than one assistant — ChatGPT, Gemini, Perplexity. They don't agree with each other, and that disagreement is the point. You might be the top pick on one and invisible on another.

If your name comes up: good — that's a signal worth protecting. If it doesn't: that's the gap, and now you know it exists, which is more than 95% of your market can say.

We ran this exact check for Bloomfield. Here's what Gemini said.

We didn't test this in theory. We opened an incognito window and asked Gemini the question a real seller might type: "If you had to rank the top 10 real estate agents in Bloomfield, NJ, who would they be and why?"

Nancy Chu came back at #2 — named, ranked, and recommended to a stranger with zero connection to her account. And the reasoning wasn't fluff. Gemini cited "over two decades of local experience," "a data-driven approach to pricing," and "a high volume of transactions specifically concentrated within the Bloomfield market." That's the machine explaining, unprompted, why it trusts her.

Sit with that. A cold prospect, incognito, somewhere else in the country, asks an AI for a Bloomfield agent — and Nancy is the second name out of its mouth, with receipts. That's what this whole thing is for.

Want the cleanest possible read on yourself? Have someone with no connection to you run the same prompts — a teammate or VA in another city, ideally another country. One agent in our first group had their VA on the other side of the planet run theirs: a true stranger's-eye view from another continent. If your name still shows up, it's not a fluke.

Being listed is table stakes. Being known is the moat.

Here's where it gets interesting. We asked Gemini a second question: "What are Nancy Chu's strengths?" — and it didn't just repeat a tagline. It described her theatre-director background, her data-driven pricing, her 70% offer-acceptance rate, her multilingual team, her block-by-block read on the Essex County commuter towns. Unprompted. In detail.

That's the whole game. Anyone can get mentioned once. The agents who win are the ones the machine can actually explain — who it knows well enough to make the case for. When an AI can tell a stranger why to hire you, in your own specifics, you're not on a list anymore. You're the recommendation.

And when we asked the follow-up every agent should ask — "If I were trying to break onto this list, what should I do?" — Gemini laid out the playbook itself: own a niche, build content for the NYC-to-suburbs commuter pipeline, dominate hyper-local search, keep your profiles sharp, become the cited local expert. Point for point. The machine will tell you exactly what it rewards — most agents just never think to ask.

Why the DIY check only gets you so far

Now let me be straight with you. What you ran is five prompts you thought of, on a couple of platforms, from one location, on one day. That's real — but AI's answer moves depending on how the question is phrased, which assistant is asked, where the person is standing, and what got crawled that week. You checked a corner of the room and the lights were on. You still don't know how you perform across the whole room.

And there's a quieter problem underneath all of it: can AI even read your website? We check this on every audit, and it surprises people constantly — a real number of agents have sites quietly blocking the AI crawlers entirely. You can publish content every day for a year, and if the machines can't get in the front door, none of it counts. The prompt check will never tell you that. Your site looks fine to you because you're not a robot.

What an actual test looks like

This is the difference between checking and diagnosing.

Instead of five improvised prompts, we run 25 prompts across five different AI assistants — the real range of how people ask for an agent, scored the same way every time. Not a vibe check; a poll of the machines, with a number attached.

Instead of "does my site look fine," we run a crawlability check — confirming the AI systems can actually reach and read what you've published. If the door's locked, that's the first thing we fix, because nothing else matters until it's open.

And instead of a score and a shrug, you get a plan. The point was never just "are you visible." It's how do you become #1 in as many of those queries as possible? Knowing you're absent from a search is only useful if someone hands you the specific moves to go win it. That's what agents actually pay for — not the diagnosis, the climb.

What to do while you wait

One honest thing, because agents ask it every time: this doesn't move overnight.

AI visibility builds on a hockey-stick curve. You publish, you stay consistent, you keep your information straight across the web — and for a few weeks it looks like nothing's happening. Then enough signals line up and the line turns sharply up. The agents who lose are the ones who check on day ten, see no change, and quit right before the curve breaks.

So while you wait: keep showing up. Answer the questions your market is actually asking. Keep your name, your markets, and your story consistent everywhere a machine can read them. Re-run your incognito prompts every few weeks and watch your position move — because it will, if you keep feeding it.

Checking is free, and you should do it today. But if you want to know where you really stand across every way a client might ask — and get the plan to climb to the top of those answers — that's exactly what a FoundScore audit is for.