AI Marketing Capabilities and Purchase Intention in Indonesian E-Commerce: Mediating Effects of Platform Trust and AI Transparency, and a Strategic Marketing and Digital Human Resource Capability Perspective

Background AI-driven personalized marketing is increasingly important in e-commerce, yet how personalization translates into purchase intention remains underexplored. Platform Trust and AI Transparency have received limited empirical attention as mediating mechanisms, particularly in the Indonesian context. Methods A quantitative cross-sectional survey was conducted from February to March 2025 among 248 active e-commerce users in Jakarta, Yogyakarta, Bandung, and Surabaya. Data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) in SmartPLS 4, with 5,000 bootstrap resamples to assess direct and mediated effects. Results The measurement model demonstrated acceptable reliability and validity (outer loadings 0.719–0.902; Cronbach’s alpha 0.757–0.882; composite reliability 0.846–0.925; AVE 0.578–0.803; HTMT <0.85). The model explained 60.4% of Purchase Intention, 46.2% of Platform Trust, and 33.2% of AI Transparency. Perceived Personalization had the strongest association with Platform Trust (beta = 0.442, p < 0.001), while its largest mediated effect on Purchase Intention operated through Platform Trust (beta = 0.125, p < 0.001). Perceived Usefulness directly predicted Purchase Intention (beta = 0.205, p < 0.001) and also showed significant indirect effects through both mediators. AI Responsiveness operated primarily through Platform Trust (beta = 0.056) and AI Transparency (beta = 0.057). All six mediation paths were significant (p < 0.001). Conclusions Within the limits of the cross-sectional design, personalization and AI responsiveness were not primarily associated with purchase intention as standalone predictors; their effects operated substantially through Platform Trust and, to a lesser extent, AI Transparency. The findings suggest that AI marketing capabilities generate commercial value when translated into relational and cognitive assets. Platforms should therefore combine personalization and responsiveness with trust-building mechanisms, interpretable recommendations, and adequate digital capabilities. Longitudinal or experimental research is needed to establish temporal and causal relationships.