Abstract for: An Integrative Framework for Agentic AI Implementation

This workshop develops the framework of an integrative approach to agentic AI integration that reflects the complex adaptive nature of organizational systems where people, processes, tacit knowledge, and culture continuously co-evolve (Heifetz et al., 2001). Current implementation approaches emphasize technical orchestration — protocols for agent coordination (LangGraph, AutoGen), data integration (Model Context Protocol), and workflow automation. While these tools address crucial technical challenges, they rest on three problematic assumptions: i) knowledge infrastructure exists and is well-maintained; ii) system dynamics are manageable through workflow design alone; and iii) human capability preservation happens as a result of adoption. The current gap between technical capability and organizational integration is not a future concern; it is causing implementation failures now. The session will be conducted as a hands-on workshop and will demonstrate how to improve the macro-cognitive understanding of workflows, people, and data interdependencies through an integrative framework that maps foundational elements, the operational practices those elements enable, and the emergence — or absence — of three capabilities: discernment, agency, and cross-boundary collaboration. The central pedagogical design is a discovery exercise: groups receive different sets of foundational elements and operational practices, map them through the framework, and arrive independently at an integration outcome — without being told in advance which pattern they are building toward. The debrief reveals that the room has produced two divergent outcomes from different starting conditions, making visible — experientially — why technically equivalent organizations integrate so differently.