Agentic AI Readiness in Italian SMEs: System Integration, Business Model Reinvention and the Intention–Capability Gap in the Post-Globalisation Era.

In the era of post-globalisation and agentic AI, SME competitiveness increasingly depends not only on adopting digital tools, but on redesigning the managerial, infrastructural and governance conditions that allow intelligent systems to act across organisational boundaries. Despite growing awareness of AI as a strategic imperative, many SMEs fail to translate this awareness into a structured, governed and integrated capability. The gap between declared intent and actual organisational capability, what we term the Intention-Capability Gap (ICG), remains large and poorly understood, yet is critical to navigate as AI transitions toward agentic architectures that demand deep system integration. This paper investigates the structural conditions that explain the ICG in Italian SMEs, using original survey data collected from 124 Piedmontese (Italy) firms in 2025. We focus on three interlocked dimensions: (1) the nature of the AI benefits that firms seek, predominantly operational efficiency, which technically requires deep system integration; (2) the actual level of digital maturity and system integration achieved; and (3) the governance and training configurations that predict readiness. Our main findings are: the primary expected benefit from AI is operational efficiency (cited by 67% of firms), yet only 16% report high system integration and 43% cite poor AI knowledge as their leading barrier. Mean ICG is +13.7 (normalised to [−100, +100]), with 68% of firms aspiring to more AI than their current infrastructure can support. Digital maturity varies significantly by firm size (Kruskal-Wallis H = 18.89, p < .001), consistent with the resource-based prediction that organisational scale is a structural precondition for digital infrastructure investment (Guangliang et al., 2025). Training intensity is the dominant predictor of both digital maturity and AI strategic awareness in multivariate analysis; firms adopting a mixed governance model display higher maturity and integration in bivariate analysis, though this advantage is largely mediated by size and training once controls are applied. An exploratory model predicting the ICG from structural firm characteristics yields no significant predictors (adj. R² = −0.010). This result is interpreted with caution, given the measurement limitations inherent in difference scores derived from single-item ordinal variables. However, it is consistent with the hypothesis that the gap reflects firm-specific dynamics, including legacy infrastructure configuration and leadership-level digital awareness, that vary independently of size, sector, and governance type. The paper argues that AI readiness in Italian SMEs is fundamentally a knowledge and capability challenge, not a financial one. Policy implications favour targeted training support and hybrid governance models over generic investment incentives, and point toward the need for firm-level diagnostic assessment of integration readiness alongside standard categorical support mechanisms. The paper contributes to the conference theme by showing that agentic business model reinvention is constrained less by tool availability than by the managerial capacity to integrate systems, develop knowledge micro-foundations and govern hybrid human–AI transformation in fragmented post-globalisation environments.