The generalist’s context advantage: why family physicians may be uniquely positioned for artificial intelligence-augmented clinical practice Authors: Sang Hyun Ahn Published: 2026 To the Editor, Artificial intelligence (AI) systems are entering clinical practice through documentation assistants, diagnostic aids, and decision support tools; however, the quality of their outputs is highly dependent on the context supplied by human users [1,2]. The field has evolved from “prompt engineering,” where precise instructions are crafted, to “context engineering,” where AI systems are systematically supplied with comprehensive, structured information [3,4]. Recent Nature Medicine perspectives have independently identified context engineering as a central professional task for physicians [5] and proposed “context switching” as the defining challenge when scaling medical AI [6]. Therefore, we propose the “generalist’s context advantage,” a conceptual framework that suggests that family physicians, by virtue of their comprehensive, longitudinal, and biopsychosocial approach, may be uniquely positioned to supply AI systems with the broadest clinical context. The daily work of a physician, which includes taking histories, conducting physical examinations, interpreting tests, and clinical reasoning, is fundamentally the work of gathering, structuring, and integrating context [2]. When a physician asks, “When did the pain start?” they are performing a temporal context injection. When they combine symptoms, signs, and laboratory values into a clinical assessment, they are performing multimodal context integration. Studies evaluating large language models for complex diagnostic cases with rich clinical narratives have reported substantial diagnostic accuracy [7,8], which suggests that the quality of human-supplied context may be a rate-limiting factor in AI-assisted reasoning. We propose that the effectiveness of physician–AI interactions may be conceptually modeled as: AI output quality ≈ f(context breadth, context depth, model capability). This heuristic illustrates directional relationships, rather than a mathematically precise formulation. Within a narrow specialty domain, deep context may compensate for limited breadth. However, as the capability of models increases and becomes commoditized, the differentiating variable shifts to human-supplied context. The four pillars of primary care by Starfield [9]—comprehensiveness, continuity, coordination, and first-contact access [10]—and the emphasis of the World Organization of Family Doctors on person-centered care map directly onto context engineering competencies [11], as shown in Table 1. Family physicians routinely integrate biological, psychological, social, and environmental dimensions across the lifespan of a patient, potentially representing the broadest context among different types of physicians. This framework does not diminish the value of subspecialty expertise. Subspecialists provide indispensable domain-specific depth. For example, the nuanced hemodynamic reasoning of a cardiologist, tumor board synthesis of an oncologist, and rapid multisystem triage of an emergency physician represent context with extraordinary clinical value. Moreover, fields such as infectious diseases and emergency medicine also manage deeply complex, multidimensional patient scenarios. The proposed advantage of the generalist does not lie in surpassing this depth; rather, it lies in the systematic breadth and routine integration of a patient’s concurrent mental health status, occupational exposures, medication adherence barriers, family dynamics, and social determinants, which may not be captured systematically during a domain-focused consultation [12]. The two forms of context—domain-specific depth and cross-domain breadth—are complementary, rather than competing. Furthermore, patients are increasingly engaging in self-directed AI use for health purposes. We have proposed the term “vibe health” to describe this phenomenon, analogous to “vibe coding” in software engineering [13]. Although this democratizes access to health information, patients generally lack the clinical training required to structure queries optimally or critically evaluate AI outputs [14,15]. Therefore, the role of a family physician in the AI era extends beyond that of a diagnostician and care coordinator to become a context engineer who structures and enriches the clinical context for the benefit of patients. In clinical practice, family physicians do not need to become AI specialists; they need to continue practicing comprehensive medicine while learning to interface with AI tools [16]. For medical education, the clinical skills already taught, such as thorough history taking, biopsychosocial assessment, and longitudinal relationship building, should be recognized as context engineering competencies in their own right. Figure 1 shows a conceptual 2×2 matrix that illustrates how the combination of a broad context and high model capability may create the optimal scenario for physician–AI interaction. This letter presents a conceptual proposal, rather than empirical evidence, and we acknowledge this as a key limitation. The “generalist’s context advantage” is a hypothesis that must be validated through comparative studies measuring AI output quality as a function of the context supplied by different types of providers. Context breadth alone does not guarantee context quality, and a thorough subspecialist working within their domain may provide more useful context than a cursory generalist. The relative importance of breadth versus depth likely varies by clinical scenario. For example, acute coronary syndrome may benefit more from deep cardiological context, whereas multimorbidity management may strongly favor comprehensive contextual input [12]. Moreover, our proposed model does not account for emerging hybrid roles, such as internists with broad training or generalist-trained subspecialists, that may also supply broad context. Therefore, we offer this framework to stimulate the empirical research required to test these propositions. In an era where the output quality of AI is increasingly determined by the richness of the supplied context, the comprehensive, longitudinal approach of family medicine may represent a competitive advantage, rather than a limitation to be overcome [17]. The competencies that define the specialty—broad history taking, integrative clinical reasoning, longitudinal care, and crossdomain coordination—may be precisely the competencies that maximize AI utility. Family physicians are potentially the most experienced broad-scope context engineers in medicine, and the barrier to AI adoption may be a recognition deficit, rather than a technology deficit.