<p>This study examines how emotional intelligence and self-efficacy shape chatbot adoption in Tunisia's tourism sector by integrating the UTAUT2 with SCT. Addressing a critical gap in AI-driven technology acceptance research, we investigate the psychological and emotional mechanisms underlying users' engagement with conversational agents in emerging markets. A sequential mixed-methods design combined semi-structured interviews with survey data (N = 301) from Tunisian tourism stakeholders, analyzed via CB-SEM. Results demonstrate that emotional intelligence dimensions sequentially activate self-efficacy (R² = .298), which significantly influences performance expectancy. Multi-group confirmatory factor analysis established metric invariance across supply-side and demand-side stakeholders. Mediation analysis confirms that self-efficacy and performance expectancy fully mediate emotional intelligence effects on adoption intentions. These findings extend technology acceptance theory by demonstrating that emotional competencies constitute foundational psychological infrastructure enabling AI adoption in service-intensive sectors. The findings offer actionable implications for enhancing digital transformation and service innovation in tourism organizations.</p>