Exploring the Impact of Talent Management Strategies on AI Adoption in Saudi Arabia’s Emerging Tech Startups: The Mediating Role of Knowledge Sharing

Abstract

Saudi Arabia's Vision 2030 emphasizes AI-driven digital transformation, yet tech startups struggle to scale AI beyond pilots. Purpose: This study examines how talent management (TM) strategies—attracting-selecting (AST), developing (DT), empowering (ET), retaining (RT), and career succession (CS)—shape AI adoption, and whether knowledge sharing (KS) mediates this relationship. Method: Using probability-based systematic random sampling of employees (n=337, N=2,308) across Saudi AI-adopting startups, the model was analyzed with PLS-SEM (SmartPLS 4). Findings: AST, DT, and ET positively affect AI adoption; RT shows no effect; CS exhibits a negative effect. KS partially mediates AST, DT, ET, and CS effects, indicating TM practices influence adoption primarily through knowledge institutionalization. Implications—Industrial: Startup leaders should integrate KS infrastructures with TM initiatives. Recommended practices: (1) cross-functional AI taskforces with rotating membership; (2) peer-learning sessions where early adopters mentor colleagues; (3) searchable repositories (wikis, Confluence) documenting implementation lessons and troubleshooting guides; (4) succession systems prioritizing collaborative knowledge transfer (mentoring, communities of practice) to prevent silos. Empirical evidence shows succession planning without KS scaffolding correlates negatively with adoption (β = -0.182, p < .01), highlighting knowledge-hoarding risks. Academic: The study extends technology-acceptance theory by integrating human-capital antecedents and positioning KS as the pivotal mediating mechanism in resource-constrained startups. Testing 16 structural paths across five TM dimensions addresses three gaps: (1) mechanistic under-specification, (2) construct aggregation bias, and (3) non-Western context neglect. The mediation framework—validated through bootstrapped indirect effects—provides a replicable blueprint for future research examining causality, moderators (industry velocity, founder literacy), and boundary conditions.