A buyer moving to your city used to open a portal. Increasingly they open ChatGPT, Perplexity or Google’s AI answers and ask “who is a good buyer agent in Oak Park?” or “what should I know about buying in this neighbourhood?” Whatever those tools say is the first impression you never got to make. Most agents have no idea what that answer is.
Run the questions your buyers actually ask
Open each assistant and ask the ten questions a real client would, in their words. “Best real estate agent in [your area]”, “who should I use to sell a condo in [neighbourhood]”, “is [your brokerage] any good”. Save the answers. That is your baseline, and it usually takes ten minutes.
Three outcomes are possible: you are named, a competitor is named, or the assistant lists directories like Zillow and Realtor.com instead of any person. The third is the most common, and it tells you the model has no strong signal about individual agents in your market, which is an opening.
What makes an assistant able to cite you
These systems quote sources they can read and trust. In practice that means a site with real text about who you serve and where, consistent details across your Google Business Profile, brokerage bio and social profiles, and third party pages that mention you in context: local press, community sites, review platforms. A one-page site with a contact form gives a model nothing to work with.
The part most agents miss
Assistants weight community discussion heavily. Local subreddits, neighbourhood forums and Facebook groups end up quoted far more often than polished marketing pages. Being genuinely useful in those places, answering questions without pitching, does more for your AI visibility than another round of listing photos.
Check again in a month
Run the same ten questions monthly and note what changed. This is the same discipline as tracking your search rankings, only the results page is a paragraph and the competition is whoever the model happened to trust. Knowing where you stand is the entire first step.