Observing the observers of projects: reconceptualizing the PMO through a von Foersterian second-order cybernetics lens

Purpose This conceptual study examines how an AI-enabled Project Management Office (A-PMO) can be understood as a second-order observing system rather than merely as an instrument that observes and controls projects. Drawing primarily on Heinz von Foerster and positioning his work in relation to management cybernetics, conversation theory, self-referential social systems and ethico-onto-epistemology, the research asks how a PMO can observe the distinctions through which human and algorithmic observers construct project-governance reality. Design/methodology/approach The research is presented as a reflexive theoretical essay rather than as an empirical study or a formal problematization analysis. Its mode of inquiry consists of: (1) reconstructing four von Foersterian distinctions; (2) re-entering those distinctions into the manuscript's own observer position; (3) contrasting them with established and critical PMO scholarship; and (4) deriving design propositions, observable indicators and possible empirical tests for an A-PMO. Findings The study develops four connected principles: non-trivial governance, eigenvalue-based recursive stability, protection of undecidable governance questions from algorithmic foreclosure and explicit observer inclusion. The five-layer A-PMO framework provides an organizational configuration through which these principles are translated into PMO practices, including decision rights logs, risk classifications, protection of minority viewpoints, accountability records, recurring service reviews and second-order audits. The central argument is that the A-PMO's task is not to eliminate governance uncertainty, but to preserve the conditions under which human decisions can still be meaningfully and accountably evaluated. Originality/value The study contributes a second-order theoretical account of the A-PMO: it reconceptualizes algorithmic project governance itself as a form of observation, thereby extending existing critical, sociotechnical and governance perspectives on PMOs with an epistemological layer that these perspectives do not yet supply.