Hari PrasathJothiswaran

Addressing the limitations of conventional Material Requirements Planning in contemporary supply chains, this paper introduces an approach centred on constrained stochastic optimization. A simulation-optimization framework minimizes inventory cost while holding service level targets and operational constraints, and returns the optimal reorder parameters that MRP systems need to run against.

This paper examines how large language models can be integrated across the BPM lifecycle — identification, discovery, analysis, redesign, implementation, monitoring — to sharpen process analysis, design, and execution. It sets out the integration points, a set of forward-looking research questions, and a structured agenda for the practical and theoretical gaps that remain.

An enhanced Variable Neighborhood Search approach is benchmarked against a genetic algorithm and an exact method on nut supply chains. VNS beats the genetic algorithm on both computational efficiency and solution quality, and trades a small amount of precision against the exact method for a large cut in run time — a workable position for agricultural networks where resources and distribution shift constantly.

Drawing on data from 360 employees at textile firms in Saudi Arabia, this study tests how digital infrastructure and management information systems strengthen supply chain management. Firms with stronger digital capability were less exposed to disruption, and risk mitigation partially mediated the path from infrastructure to robustness — pointing investment toward both the systems and the mitigation practices around them.

Deterministic assumptions and static heuristics break against real-world variation. This paper applies Generative Probabilistic Planning to produce alternative supply chain plans that carry uncertainty with them, using attention-based graph neural networks alongside offline reinforcement learning and probabilistic simulation. Tested on data from a multinational firm, the approach measurably narrows the gap between supply and demand.

What do BPM practitioners actually think about putting autonomous agents inside their processes? This paper works through three dimensions — agent autonomy, governance mechanisms, and human-agent collaboration — and reports both the efficiency and decision-making gains people expect and the risks they name around bias, transparency, and organizational change.

Continuous control models are used to allocate, distribute, and store materials across a supply network without a planner in the loop. The method targets efficiency and profitability on exactly the dynamic decisions that defeat static rules, and is evaluated against traditional supply chain management techniques.

Helpdesk incidents on a supply chain platform are resolved through contextual vector retrieval: a vector database paired with an LLM that pulls new information at query time rather than relying on what it was trained on. The retrieval model is built for logistics and procurement problems specifically, and a real incident study shows the gains in resolution efficiency, throughput, and cost.