Artificial intelligence in public procurement: a multilevel dynamic capabilities perspective

Purpose This study explores how artificial intelligence (AI) adoption in public procurement emerges through dynamic capabilities across individual and organisational levels. Design/methodology/approach A qualitative study was conducted with 20 professionals from 13 Finnish public organisations engaged in AI-related procurement initiatives. Using the Gioia methodology, interviews and supplementary documents were analysed to identify enablers, barriers and transformation mechanisms. Findings The findings show that dynamic capabilities – sensing, seizing and transforming – are distributed across levels and shaped by bottom-up and top-down mechanisms. Individual experimentation feeds organisational learning, while leadership and governance structures enable or constrain scaling. Most organisations remain in the early phases, with transformation hampered by silos, vendor lock-in and weak orchestration. Originality/value The study advances dynamic capabilities theory by demonstrating its multilevel nature in the public sector. It provides actionable insights into aligning individual initiatives with organisational structures to achieve sustainable, AI-enabled procurement transformation.