Local AI search is what happens when a patient asks an AI assistant for nearby care: "eye doctor near me", "pediatrician open Saturday in Mesa", "best physiotherapist near Union Station". Instead of a map and ten links, they get a short written answer naming a few practices. The term itself barely registers as a search ("local ai search" gets about 10 searches a month and "ai local seo" 40), but the patient-side demand it touches is enormous: "eye doctor near me" gets 450,000 searches a month and "pediatrician near me" 246,000 (Google Ads Keyword Planner data via DataForSEO, US, 12-month average to Sep 2026). Every one of those can now become an AI answer.
In short: Local AI search is an assistant answering a "near me" health question with a short list of practices. Assistants estimate where the patient is, search listings, websites and reviews, and write an answer. That's why two patients can get different names. A practice improves its odds with complete, consistent listings, a page per service and steady reviews.
What local AI search is
It's the meeting point of two things patients already do: local search and asking an assistant. Three surfaces matter:
- Google AI Overviews and AI Mode, which sit inside Google Search and draw on Google's index and local business information.
- Chat assistants with web search: ChatGPT, Gemini, Claude and Perplexity.
- Follow-up conversations, where the patient narrows the list: "which of those take Medicaid?", "which is closest to the 101?"
What makes it different from a map search is the filter in the middle. A model decides which practices to mention and what to say about them, based on whatever sources it found and trusts.
To see whether Claude and Gemini name your practice for local questions today, run the free marketing audit.
How assistants work out where the patient is
Location is the first input, and providers document it unevenly:
| Assistant | What the provider documents about location |
|---|---|
| Gemini app | By default uses the general area from your IP address or the Home and Work addresses in your Google Account; uses precise device location with your permission (Google (opens in a new tab)) |
| Claude | "May use your location (inferred from your IP address) when responding to a request for localized results" (Anthropic (opens in a new tab)) |
| ChatGPT | OpenAI's developer docs show its web search can take an approximate user location (city, region, country, timezone) (OpenAI (opens in a new tab)); how the ChatGPT app sets it isn't confirmed here |
| Google AI Overviews and AI Mode | The help pages don't describe location handling; AI Mode can personalize using past searches for signed-in adults with Search history on (Google (opens in a new tab)) |
| Perplexity | Not documented |
Distance tolerance also depends on the kind of care. Google's Search Quality Rater Guidelines note that people "might be willing to travel a little farther" for doctors' offices, and may accept results even farther away for "a very specialized medical clinic" (Google (opens in a new tab)). A pediatric practice and a Mohs surgeon draw from different radii.
Listings, maps and reviews as inputs
Google documents the most:
- Place cards in AI Mode. Google said in May 2025 that for local spots, AI Mode lets you "quickly see info like ratings, reviews, and opening hours", and call or get directions from the answer (Google (opens in a new tab)).
- Business Profiles feed AI answers. Google's generative AI guide says local business information can appear in AI responses and that Business Profiles can help your services be visible there (Google Search Central (opens in a new tab)).
- Local ranking basics. Google says local results are based mainly on relevance, distance and how well-known a business is, that "more reviews and positive ratings can help your business's local ranking," and that there's no way to pay for a better local ranking (Google Business Profile Help (opens in a new tab)).
The other assistants don't say which listing, map or review providers they use. When you test local health questions yourself, expect to see directory and review pages among the citations; that's an observation to verify for your city, not documented behavior. Treat every public listing as a possible input.
Why two patients get different answers
Two people asking "best pediatrician near me" five minutes apart can see different practices. The usual reasons:
- Location. IP-based location can be off by a town, and a phone with precise location on narrows the answer further.
- Phrasing. "Near me", "in Mesa" and "near Dobson Ranch" are three different searches.
- Personalization. Google's AI Mode can draw on a signed-in adult's previous searches.
- Timing. Assistants search live, so a new review, a changed listing or a recrawled page can change the answer from one day to the next.
- The model. Providers update models, and some route questions to different models.
- Conversation context. A follow-up question inherits the first one's assumptions.
- Error. Google's help page says AI Overviews "can and will make mistakes" (Google Search Help (opens in a new tab)), and so can every other assistant.
This is why nobody can promise a mention. AI answers vary by phrasing, location and day.
What a practice can do: the local signals checklist
This isn't a local SEO tutorial; the linked guides cover each step. It's the list of signals that local answers draw on:
- Google Business Profile complete and verified: primary and secondary categories, services, hours (including holiday hours), attributes and photos. See Google Business Profile for doctors.
- Bing Places profile claimed and matching your Google profile. See Bing Places for practices.
- Apple Business Connect claimed, so Apple Maps shows the same details. See Apple Business Connect.
- Identical name, address and phone on your website, insurer directories and health directories.
- A page per service on your website, naming the city and neighborhoods you serve.
- A page per location if you have more than one office, each with its own hours and phone.
- Recent, steady reviews, with replies that never confirm someone is a patient.
- Search crawlers allowed, so assistants can read the pages behind your listings.
Checking local AI answers across your service area
Testing from your front desk shows one view: the one tied to your office's internet connection. Patients ask from everywhere else.
- Map your service area. Pick three to five places patients come from: your neighborhood, two suburbs, and the area near your hospital affiliation.
- Write place-named prompts for each: "pediatrician in Gilbert", "pediatrician near Dobson Ranch", "pediatrician in Mesa that takes AHCCCS".
- Run a "near me" prompt too, and note which city the assistant seems to assume.
- Log by place: who is named, in what order, and which sources are cited.
- Repeat weekly and look for places where you consistently drop out.
Illustrative example: A fictional practice, Desert Sky Pediatrics in Mesa, is named for "pediatrician in Mesa" in all four weekly checks but never for "pediatrician in Gilbert", though a third of its patients live there. Its service page mentions only Mesa. The fix is on the page and the profile, not in a new tactic.
For how one assistant handles these questions in detail, see how ChatGPT recommends practices. Google's conversational mode has its own quirks, covered in Google AI Mode. The full picture is in AI visibility for practices.
Where Medfy fits
Medfy's AI visibility tracking runs weekly checks of ChatGPT, Gemini, Claude, Perplexity and Google AI Overviews using the questions patients ask (e.g. "best family doctor near me" in your city) and shows whether you're named, who is recommended instead and how that changes. Medfy's first-party analytics counts visits from AI assistants. The free marketing audit includes an AI visibility score that asks Claude and Gemini (with live web search) whether they name the practice and checks that the crawlers behind ChatGPT, Perplexity, Claude and Gemini can read the site.
Frequently asked questions
Do AI assistants use Google Maps data?
Google's own AI features do draw on local business information: AI Mode shows place cards with ratings, reviews and opening hours, and Google says Business Profiles can help businesses appear in AI responses. ChatGPT, Claude and Perplexity don't publish which map or listing providers they use. Keep your Google, Bing and Apple listings identical so every assistant finds the same facts.
Why does ChatGPT show different practices to different people?
Because each answer is built fresh from where the person seems to be, how they phrased the question and what a live search returns at that moment. OpenAI's developer documentation shows its web search can take an approximate location such as city and region. Model updates and conversation context add more variation, so judge your visibility over weeks of checks, not one answer.
Do reviews affect AI recommendations?
For Google, yes in part: Google says more reviews and positive ratings can help local ranking, and AI Mode shows ratings and reviews on place cards. Other assistants don't document how they weigh reviews, but review sites are among the pages they can find and cite. Steady, genuine reviews and replies that never confirm someone is a patient are worth the effort.
How do I check AI answers for my city?
Pick three to five places in your service area and write a prompt naming each, such as "pediatrician in Gilbert". Run them in ChatGPT, Gemini, Claude, Perplexity and Google weekly, logged out where possible, plus one "near me" prompt. Record who is named, the order and the sources cited, and look for places where you consistently drop out.