Centaurization, an Operational Framework for Structuring Human–AI Collaboration in Expert Professional Disciplines: Evidence from a Business Continuity Management Case Study

Integrating artificial intelligence into the work of an expert practitioner is not merely a matter of tool selection: it requires knowing precisely which activities to entrust to the AI, which to reserve for the human, and how to faithfully translate that partition into a real collaboration specification. Existing frameworks address individual dimensions of human–AI collaboration — including task typologies, performance evaluation, or system design. This paper introduces an integrated, operational protocol spanning activity mapping, task allocation, implementation specification, and post-deployment recalibration. It proposes the Centaurization Methodology, a Design Science Research (DSR) artifact that structures human–AI collaboration in four sequential phases: Diagnosis (field activity mapping + Practitioner Verdict), Binarization (D/M/S classification via sequential binary tree + transversal rules), Instantiation (compatibility test + technology-agnostic Instantiation Profile + Practitioner Visa), and Feedback (Residual Arbitration Rate + continuous recalibration). The framework proposes five axes designed to structure the work of any expert practitioner — ASSESS, DESIGN, DEVELOP, INTERVENE, PROVE — applicable independently of the discipline or reference standard. A complete (Phases I, II and III) case study has been conducted on Business Continuity Management, producing 31 practitioner activity sheets and 104 binarized tasks (practitioner-validated by two certified BCM experts), with a M-ratio (proportion of Mixed tasks) of 53%. Two additional disciplines (Risk Management, Information Security) have completed Phase I Step 1 (documentary pre-mapping) and await practitioner validation.