Responsible Automation in E-business: A Framework for Agentic AI Integration and Governance

Abstract

This study proposes the Responsible Autonomy Framework (RAF), a layered governance architecture for agentic artificial intelligence (AI) in e-business. The RAF integrates operational efficiency, strategic adaptability, and responsible oversight into a single layered model. Using a conceptual research design grounded in an integrative literature review, agency theory, and sociotechnical systems thinking, the study develops a three-layer model comprising operational autonomy, strategic autonomy, and responsible autonomy. In which responsibility acts as the governing layer rather than an afterthought. The review finds that existing governance frameworks address transparency, accountability, or fairness largely in isolation, with none spanning the full path from operational automation to embedded governance, while the RAF addresses this by treating responsible design as a condition for sustainable autonomy rather than a competing value to be balanced against it. The framework offers managers actionable guidance for deploying agentic AI responsibly and offers regulators a structured basis for balancing innovation with oversight, while extending agency theory and sociotechnical systems theory through a reconceptualisation of agentic AI as a sociotechnical principal-cum-agent.