Beyond Assistance: A Role-Based Conceptual Framework for Generative and Agentic Artificial Intelligence in Case-Based Management and Governance Research

Abstract

The use of artificial intelligence (AI) in qualitative management research has so far been framed predominantly as assistance: a researcher interacting with generative AI during the design or analysis of a case study. Yet methodological discussions still offer limited understanding of how qualitatively different forms of AI involvement reconfigure the production of theoretical insight in case-based research. The existing differentiation remains stage-bound and tool-centred; what is missing is a systematic coupling of AI roles to forms of theorising and to specific validity threats. Drawing on a review of 139 documents, the authors develop a four-role conceptual framework, AI as Discoverer, Selector, Co-Constructor and Generator. Each role is examined through affordance theory, theorising from cases and sociomaterial perspectives in order to analyse how different forms of AI involvement influence theorising, interpretation and engagement with empirical material. The study's contribution lies less in naming four roles than in coupling them to forms of theorising and to role-specific validity threats. AI involvement cannot be treated as a uniform methodological condition: different roles raise different implications for contextual understanding, methodological rigor, researcher oversight and the empirical status of research material. We advance six propositions linking each AI role to conditions of legitimacy, characteristic validity threats and disclosure requirements, and we illustrate the framework through a corporate-governance research scenario. The paper closes with role-sensitive guidance for authors, reviewers and editors of governance and R&D case research.