AGENTONOMICS: The design principle for autonomous AI agents to achieve economic growth

Abstract

Agent-based artificial intelligence is shifting the focus of management: models that generate answers are giving way to systems that pursue objectives, utilise tools, communicate with other actors and act within set boundaries. This paper presents Agentonomics as a proposed management discipline for this development. The approach combines the St. Gallen concept of integrated management with the computer science of cooperating AI agents and distinguishes between three levels. The Agentonomics concept provides the normative and cognitive ‘DNA’: organisations are understood as an interplay of external, business and management systems; the central task is dynamic coherence control, supported by the recursive 5C Navigator. Agentonomics theory describes AI agents as functional economic actors within a polycentric economy of capabilities, in which claims to ownership, control and participation in data and transferable competencies must be reorganised. The Agentonomics model operationalises these ideas in the six-stage Design & Management Reference Framework (AD- MRF), in management agent systems and in a leadership cockpit. Here, internal and external evidence, scenarios, functional role agents, human interventions, implementation and learning are combined to form a transparent leadership control loop. This paper outlines practical fields of application, the evolution from assistant to functional AI CEO, and a research agenda. Agentonomics is understood not as a theory that has already been empirically validated, but as an ideal-typical research and design programme, the viability of which must be system- atically tested in teaching, corporate laboratories and real-world decision-making situations.