Implementing Artificial Intelligence in Data-Driven Enterprises through a Ten-Phase Framework Based on Multiple Case Studies

Purpose. To develop and empirically validate a ten-phase framework for the implementation of artificial intelligence (AI) in data-driven enterprises, conceptualizing AI adoption as a socio-technical transformation integrating strategic, technological, and human dimensions. Method. Multiple case study design combined with Design Science Research. Data were collected from twelve enterprises across manufacturing, logistics, and healthcare sectors through interviews, workshops, process datasets, and surveys. Cross-case and within-case analyses were used to validate and refine the framework. Findings. Organizations that completed all ten phases recorded measurable gains: +14% operational efficiency (OEE); –32% decision latency; +11% first-pass yield (FPY); >80% user adoption of AI-supported systems. Success depended less on algorithms and more on data integrity, strategic alignment, human participation, and embedded governance. Theoretical implications: Integrates fragmented models (CRISP-DM, AI Maturity Models, TOE) into a unified, process-oriented framework; redefines AI maturity as a dynamic capability-building process; extends socio-technical systems theory by proving that human participation accelerates AI adoption. Practical implications. Provides a replicable roadmap for scaling trustworthy and regulation-compliant AI aligned with the EU AI Act and ISO/IEC 42001. Demonstrates that data governance, competence development, and early compliance reduce failure risk and long-term cost. Value. The first empirically validated end-to-end managerial framework linking strategy, data governance, human competence, and regulatory accountability into a single model for AI transformation. Bridges the gap between AI research and enterprise-level implementation.