A Cross-Disciplinary Knowledge Management Framework for Artificial Intelligence in Business Projects





knowledge management, project team competencies, machine learning projects


This paper presents a knowledge management framework for designing, monitoring, and optimizing the business value of intelligent services using best practices from both management and machine learning engineering areas. The increasing interest in artificial intelligence has highlighted the difficulty of designing an appropriate business case and monitoring the real business value generated by such services due to the complexity of machine learning methods and techniques. Managers have proven frameworks for constructing key performance indicators, machine learning engineers have developed advanced methods and techniques for evaluating and monitoring Machine Learning (ML) models, and scientists have tested methods for empirically evaluating the added value of innovations. By leveraging the strengths of each discipline, the proposed framework enables managers, machine learning engineers, and scientists to efficiently work together and optimize the value of intelligent services during their lifecycle. We propose processes and procedures for creating, capturing, organizing, storing, sharing, and using knowledge in advanced machine learning projects. He also presents key machine learning methods and techniques for monitoring and optimizing the value of intelligent services over their lifecycle, including the adaptation of MLOps methodology for continuous monitoring, reinforcement learning for continuous improvement, and CausalML methods for identifying the root causes of changes in the business value. These methods and techniques support knowledge management activities and help formulate a competency framework for team members and project stakeholders. We point out the potential of academic researchers and external advisors as catalyzers in such projects based on real-life implementation. He also proposes a method for designing the ML knowledge flywheel to ensure continuous knowledge transfer and improvement in the business-engineer-academy triangle. The approach is illustrated by a case study of the implementation of a marketing communication optimization system in a large, multinational financial company for more than 20 thousand customers in two European countries. Managers and machine learning engineers can implement the proposed knowledge management framework in various organizations for the efficient design, monitoring, and optimization of the business value of intelligent services.