A patient in Columbus opens ChatGPT and types that they need a family doctor nearby who's accepting new patients and takes a specific insurance plan. Three seconds later, the AI names two practices. One is a sprawling regional health system with a marketing budget larger than most independent clinics' annual revenue. The other is a small family medicine group that nobody in the city's marketing agencies has ever heard of. How did the small practice land in the same sentence as the health system giant?
That question sits at the center of the single most competitive category in medical AI search. Primary care is the top of every patient acquisition funnel — the specialty patients search for most, switch most often, and rely on AI recommendations for most heavily. It's also the hardest category for an independent practice to win, because the deck is stacked in ways that aren't obvious until you understand how AI assistants actually decide who to name.
Why primary care is the toughest medical AI category
Every specialty has competition, but primary care has several things working against the independent physician at once. Query volume is enormous — searches for a family doctor or an internist accepting new patients are among the highest-frequency health queries fed into every major assistant, and high volume means high competition. Primary care is also where health systems concentrate their brand-building most heavily: a hospital network doesn't have one location, it has dozens, all sharing the same entity authority, the same domain, the same aggregated review signals, and the same relentless content output. When an AI assistant weighs which entity is most established and trustworthy, the health system wins that particular contest almost every time.
But here's the part most practices miss: AI assistants default to entity authority only when nothing else differentiates the options. Brand size is the tiebreaker, not the whole game. The independent practices that break through do it by giving the AI something more specific to grab onto than "big and well-known."
Compete on specificity, not authority you can't win
A small independent group will never accumulate the domain strength, citation volume, or aggregate review count of a major regional system, and spending years chasing that is a losing game. The more winnable path is attribute specificity — the concrete, checkable details a model uses to match a narrow query to a narrow answer. Health system pages, written to serve dozens of locations at once, are structurally incapable of making narrow claims. That's the opening: same-day availability stated explicitly rather than implied, appointment lengths described honestly, direct physician access by phone or message instead of a call center, language capabilities named specifically, a direct-primary-care model explained in plain terms if that's what you offer. Every one of these is something a health system marketing page underinvests in, because a system optimized for scale can't credibly promise a personal experience. When a patient asks for a primary care doctor who actually spends time with them, the AI needs a practice that has said so in plain language — and if your site says it and the health system's doesn't, you get named regardless of who's bigger.
For a deeper look at the shared traits behind these mentions, see what top AI-recommended physicians have in common.
The insurance-completeness advantage
One of the highest-intent primary care queries is some version of asking for a doctor who takes a specific plan — a patient ready to book, filtered by the one factor that matters most to them. Here's what most independent practices get wrong: they list insurance acceptance vaguely, as "most major plans," or bury it, or leave it off entirely. That phrasing is functionally invisible to AI. An assistant trying to confirm whether a specific plan is accepted can't infer a match from "most major plans" — it moves on to a practice that names the plan directly. The fix is to list every accepted carrier by name on the website, mirror that exact list on ZocDoc (a primary source AI draws from for insurance-filtered queries), and keep both in sync, because a plan you accept in real life but omit online is a plan's worth of patients you never appear for. A practice that names all of its accepted carriers will surface for each of those queries individually, while the health system down the street hiding its plans behind a "verify your coverage" portal surfaces for none of them.
New-patient availability functions as a filter, not just a claim
The phrase "accepting new patients" is doing more work in AI search than almost any other few words in primary care, because a huge share of queries in this category carry intent to establish care — someone who moved, whose old doctor retired, or who's simply never had a PCP. AI assistants treat new-patient availability as close to a hard filter: if they can't confirm it, they won't recommend you for a new-patient query even if you'd be a perfect fit. A clear, prominent statement on the website matters, but the ZocDoc calendar is where practices quietly sabotage themselves — a calendar showing no availability for weeks, or one that hasn't been touched in months, reads to the AI as "not accepting patients" and the recommendation goes elsewhere. Treat that calendar like a storefront window: if it looks empty or dusty, both patients and the AI recommending them assume you're closed. To understand the fuller journey behind these queries, read how patients find doctors through AI assistants.
Direct primary care is a wide-open lane
Direct primary care may be the single most underexploited AI visibility opportunity in the entire specialty. A flat monthly fee, no insurance billing, direct physician access — that model occupies query territory a conventional health system literally can't enter, because searches for affordable care without insurance or a concierge-style doctor at a flat monthly rate simply don't match how a hospital system markets itself. The mistake DPC practices make is assuming patients already understand the model; they don't, and neither does the AI, unless the practice spells it out in plain language — what the model actually is, what it costs in real dollar terms rather than "contact us," what's included, and who it tends to suit best. A DPC practice that publishes that explanation clearly will surface for exactly the queries no insurance-based practice can touch, in a field that's wide open mainly because most DPC practices haven't written the content yet.
What this looks like side by side
Picture two internal medicine practices in the same neighborhood, both independent, both good. One has a generic website — welcoming copy, a vague insurance claim, a ZocDoc calendar untouched since spring. When a patient asks for a nearby internist taking new patients on a specific plan, this practice is invisible; the AI can't confirm a single relevant attribute. The other lists its accepted plans by name, states clearly that it's accepting new patients as of this month, advertises same-day sick visits and real appointment lengths, notes a language capability, and keeps its ZocDoc calendar genuinely current. Ask any of the seven assistants the same question, and the second practice gets named — often ahead of the regional health system nearby — because it matched the query precisely while the system only matched it broadly. Same neighborhood, same quality of medicine, wildly different visibility. The difference is entirely in what's documented and how specifically.
Find out where your practice stands
Primary care is winnable for independent physicians, but only for the ones who stop trying to out-brand the health systems and start out-specifying them. RankCommander shows you exactly how the seven major AI assistants currently describe your practice, which queries you're missing, and where a health system is beating you on nothing more than documentation. Run a free AI visibility scan to see your current standing, and explore our medical AI visibility platform to start closing the gap on the practices getting all the mentions.