Governing AI with Kaizen: technostress and continuous improvement in Italian public organizations

Purpose The growing integration of artificial intelligence (AI) into Italian public administration raises a central question: can person-centred human resource management practices turn technology from a source of psychosocial strain into an enabler of continuous improvement? This paper explores how formalized and participatory governance of AI adoption, structured around three Kaizen principles, is associated with employee participation in digital change and with the technostress reported by public employees. Design/methodology/approach A theory-informed qualitative multiple-case study was conducted in five organizational cases, namely Italian public administrations engaged in digital transformation projects funded by the Piano Nazionale di Ripresa e Resilienza (PNRR). In each organization, a manager–officer dyad (10 participants in total) took part in narrative interviews held between October and November 2025, focused on lived experiences of AI adoption, perceptions of human resource management and professional engagement. Transcripts were analysed through deductive, theory-driven content analysis based on the three Kaizen principles (process orientation, step-by-step improvement and human capital orientation) and on the technostress dimensions, followed by within-case and cross-case comparison. Findings Across 10 participant accounts embedded within five organizational cases, practices consistent with Kaizen were associated, within the accounts, with distinct narrative profiles of technostress and engagement. Accounts reporting all three principles described active co-authorship, professional empowerment and emerging roles; accounts reporting none described persistent techno-anxiety, disengagement and eroding professional identity. Among the creators, techno-insecurity is the one most closely tied to how adoption is governed, and is typically accompanied by techno-anxiety as an affective strain response, while overload, invasion, complexity and uncertainty appear in association with specific operating conditions. Intermediate accounts varied with the missing principle and its hierarchical level. Originality/value This paper develops the joint Kaizen–STS framework as an original analytical lens for AI governance in public administration. It provides qualitative evidence that Kaizen-consistent HR governance is associated with less salient technostress, particularly in the governance-related dimensions emerging from the analysis, and with stronger reported participation, professional agency and more favourable organizational conditions for continuous improvement. The study therefore contributes to organization and human resource management research by proposing the vertical governance gap between managerial design and operational experience as a sensitizing concept and offers public managers a diagnostic framework for PNRR-funded digital transformation across heterogeneous public-sector settings.