From Manual to Intelligent: Transforming Inventory Planning at Jazz

Abstract

Jazz operates in a very complex telecommunication environment consisting of network infrastructure in different regions, and supply chain and inventory systems. One of the most critical processes in this ecosystem that can have a direct impact on the continuity of services, reliability of operations, and finances is the inventory planning. This process can be termed as requirement forecasting, stock control, supplier lead time management and balancing of service requirements with working capital position. Although enterprise systems are now being used as part of Jazz's digital transformation, the inventory planning is still not very efficient, partly because of planners’ lack of familiarity with the systems, and the necessity to make manual checks and re-checks before finalizing the plan, according to their experience, or to react to changes in supply and/or demand. To improve efficiency, Jazz will be including AI in inventory planning, the ability to plan more efficiently based on AI forecasts, automated and intelligent decision support and improved responsiveness to demand variability. However, explainability is one of the challenges that stands large. In the traditional AI approaches, they are essentially black boxes recommending but there is no communication to the traditionally offered solution: What are the drivers for the solution and how much uncertainty does it have and which local factors may not be aware of some trade-off of some sort. This low level of explainability could cause hesitation or delay amongst decision makers and perhaps even overrides, lower levels of trust and underuse of the capabilities of AI. In the absence of transparency and systematic reasoning, AI will not be able to work as a trusted partner in making decisions. Planner need to understand how AI works, and understand why it makes certain decisions, so they can effectively intervene. Without that understanding and confidence, the effective partnership between human and AI cannot be realized. (Kai Hoberg, 2023 )