From reactive to anticipatory: a conceptual framework for agentic AI governance in smart tourism cities

Urban tourism cities face a governance crisis – the growing inability of existing institutional arrangements to reconcile visitor volumes with heritage protection and resident well-being – as visitor numbers overwhelm historic centres, heritage zones, and neighbourhoods. AI has been adopted mainly reactively – analysing past data to support decisions rather than managing flows in real time. This paper argues for a shift to agentic AI with autonomous perception, reasoning, and action. Despite advances, no governance framework exists for its use in tourism cities, where authority is fragmented among the public, private, residents, and visitors. It introduces the A2-GOVERN framework, a three-part architecture designed for smart tourism cities, and the TAFR Governance Standard – Transparency, Accountability, Fairness, Reversibility – for ethical AI governance. Using Institutional and Complexity Theory, an exploratory fsQCA analysis of 28 smart tourism cities identifies four governance configurations associated with high sustainability, showing that governance embeddedness, not technical skill, is crucial for sustainable AI-driven tourism.