AI Agent Adoption in Banking_A Quantitative Analysis with Focus on Bangladeshi Customer Services

This study examines determinants of AI agent adoption in banking through secondary data analysis of 35 Bangladeshi commercial banks over 2020–2024, complemented by global comparative benchmarking. Using panel data regression and the Technology-Organization-Environment (TOE) framework, we analyze 175 bank-year observations to test five hypotheses on adoption drivers. Results indicate cost-to-income ratio (β = -0.124, p < 0.01), organizational size (β = 0.782, p < 0.001), and digital maturity (β = 0.095, p < 0.001) significantly predict AI adoption intention, explaining 68% of variance (Pseudo R² = 0.68). Financial modeling projects sector-wide cost savings of BDT 1,600-2,300 crore annually with 14-21-month payback periods. However, infrastructure gaps (72% vs. 94% 4G coverage compared to developed markets) and human capital constraints (60–70% specialist shortage) moderate implementation feasibility. For Bangladeshi banks, findings suggest phased implementation prioritizing high-volume, low-complexity use cases yields optimal ROI (141–261% over three years). This study contributes to emerging market technology adoption literature by demonstrating cost drivers dominate in resource-constrained environments (r = 0.78), contrasting with competitive pressure primacy in developed markets (r = 0.68). Theoretical contributions include extending TOE framework application to South Asian contexts and introducing digital maturity as a significant mediating variable in AI adoption pathways.