The trend: AI could perform elements of about one-fifth of US outpatient care, according to a new McKinsey analysis. McKinsey estimates that AI could handle 16% to 22% of US outpatient claims, primarily reasoning- and dialogue-based clinical care that doesn’t require hands-on examination or treatment. This includes patient intake, triage, diagnosis, guidance, referrals, and follow-up management.

The analysis used claims data across commercial insurance, Medicare, and Medicaid to identify services that could potentially be performed by clinical AI. Across care settings, McKinsey found that physician offices and clinics, virtual care, and urgent care could see the greatest AI-driven changes.

Why it matters: AI could reduce the number of routine interactions competing for clinicians' time. An “AI front door” to healthcare would not replace doctors, but rather assess symptoms, collect patient information, and route higher-need cases to the right provider. That could leave physicians with more time for patients who need physical examinations, procedures, or complex clinical judgment. Half of physicians say AI has already increased their capacity to see more patients, per an April Philips report of worldwide clinicians.

Added physician capacity matters, as shortages compound already limited patient access in the US. More than 100 million Americans lack adequate access to regular primary care, and the US could face a shortage of as many as 86,000 physicians by 2036. Meanwhile, patients wait an average of 31 days for a new physician appointment, per a 2025 AMN Healthcare report.

Several companies offer AI for outpatient tasks, including symptom assessment and clinical triage, per McKinsey. These tools typically collect symptoms through an initial intake, then guide patients to the appropriate level of care or next steps, including connecting them with a partner health system when further care is needed. Some are even embedded directly into provider websites or patient portals. However, other AI health products go further, branding themselves as “AI doctors,” potentially implying their tech can replace a physician’s advice.

Implications for health systems and AI companies: McKinsey emphasizes that AI cannot fully substitute physicians. If AI can safely resolve some nonemergent patient needs without a physician encounter, it could expand the capacity of the existing workforce. AI-assisted diagnosis and treatment, however, face greater resistance from healthcare organizations and physicians. Medical professionals have raised concerns about such uses, including a Utah pilot allowing AI to automate certain prescription renewals.

As AI becomes a new front door to healthcare, tech companies will expand tools that assess and direct patients to care, while health systems build AI into websites, apps, and portals to compete for that first interaction. That puts AI upstream in the patient journey, influencing whether consumers seek care, what type they need, and where they go. Health systems have an incentive to build their own AI patient navigation tools so they can keep that initial interaction—and subsequent care—within their own ecosystem.