Purpose This conceptual article proposes and explores Agent Project Management (APM) as a governance-first model for integrating AI agents into project workflows. We examine how AI agents can revolutionise project management by boosting efficiency and enhancing decision-making, while addressing critical governance, accountability and risk concerns. Design/methodology/approach This viewpoint article synthesises existing research to develop a novel APM model. It is framed through an original integration of three complementary lenses: socio-technical systems, governance and decision rights, and project-as-practice. Findings The article presents the APM model, built upon four core constructs (data maturity; transparency/auditability; conditional autonomy with escalation; benefits-aligned objectives), and formulates four testable propositions linking these constructs to project performance, trust, and value. Originality/value The article's primary contribution is the development and positioning of the APM model, which provides a structured framework for human–agent teamwork with conditional autonomy and inspectable accountability. The integration of the three theoretical lenses and the articulation of the four propositions offer distinct and actionable foundations for future research and practice.
Hughes et al. (Thu,) studied this question.