A GenAI-Enabled Adaptive Framework for Demand Forecasting and Supply Chain-Aware Market Segmentation in Digital Markets
Digital retailers often promise service levels that fulfillment cannot meet. Marketing commits to a demand forecast. Operations holds inventory and capacity limits that the forecast does not see. The two functions run on different information at different speeds, and live supply limits do not reach customer prioritization before commitments go out. This paper proposes a framework to close that gap. A large language model (LLM) orchestration layer sits between demand forecasting and supply chain optimization. The paper gives a joint decision model with a utility objective and a greedy allocation algorithm, a specified LLM architecture with JSON schemas and a versioned prompt registry, and a three-tier human approval process. At the center is the Demand-Supply Segmentation Matrix (DSSM), which sorts customer cohorts into four segments using live scores for demand stability and fulfillment sensitivity. In a simulation of 50 scenarios, the framework raised service level attainment by 18 points, cut inventory cost by 24 percent, and shortened disruption recovery by 54 percent against a quarterly-review baseline. The allocation in these runs comes from the greedy algorithm, so the gains reflect dynamic, supply-aware segmentation and do not isolate the LLM’s contribution. The framework has not run in production.
Authors
- Shubham Kumbhalkar
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-15
- DOI
- https://doi.org/10.5281/zenodo.24257735
- Primary Topic
- Supply Chain and Inventory Management
- Type
- article
- Field-Weighted Citation Impact
- 0.00