Generative artificial intelligence has reached unprecedented levels of consumer familiarity, with ChatGPT alone exceeding 800 million weekly active users by late 2025. Yet a parallel body of evidence indicates that this familiarity rarely translates into measurable business outcomes for the small and medium-sized enterprises (SMBs) that adopt the technology. McKinsey's 2025 global survey found that more than 80% of organizations report no tangible enterprise-level financial impact from generative AI; the Boston Consulting Group reports that only 26% of companies have moved beyond proofs of concept; and an MIT NANDA analysis of 300 enterprise deployments found that 95% of generative AI pilots produced no measurable profit-and-loss impact. This paper argues that the disconnect is best understood as an instance of the knowing-doing gap originally described by Pfeffer and Sutton (2000), now operating at the level of the individual owner-operator rather than the corporation. Synthesizing evidence from organizational theory, transfer-of-training research, gen-AI productivity studies, and recent metacognition research, the paper proposes that the binding constraint for SMB AI adoption is not access to the tool but the application of business judgment — a tacit, context-bound layer that ChatGPT cannot supply on its own. The Agentes Para Tu Negocio model is offered as one implementation framework that operationalizes this layer through bottleneck-first diagnosis and assisted system construction, with particular relevance for Spanish-speaking owner-operated SMBs in Latin America and the United States.
Humberto Inciarte (2026) studied this question.