International Journal of Production Research

The Agentic Supply Chain: Entering a new era in AI in Supply Chain Management

Recent advances in Agentic Large Language Models (LLMs) are reshaping the discourse on how autonomous decision systems might operate in complex environments such as supply chains. The motivation for the special issue is to facilitate rigorous research in the transformative potential of agentic technology for Supply Chain Management (SCM). A key objective is to foster collaboration and unlock synergies by merging diverse perspectives from operations and supply chain management, AI, complexity science, and industrial engineering.

We characterise an AI agent as one that has a predefined or emergent goal and uses cognitive tools to achieve its objectives, such as situational awareness, data-driven learning and prediction, decision optimisation, negotiation, and problem-solving with human or other AI agents. Recognising their potential for improved decision-making, supply chain management scholars have long studied agent-based systems, especially multi-agent systems (MAS) since the 2000s (Xu et al, 2021). However, research has stalled as MAS were slow to develop, difficult to code, debug and scale, and were prone to severe coordination problems. Most supply chain agent research was not adopted in the industry.

Recent advances in LLMs have now brought in a new paradigm that might reverse this trend: LLM agents (also termed as agentic LLMs, or agentic AI) are widely expected to usher in a new era where specialist programming knowledge is no longer required (Bubeck et al. 2023). LLM agents utilize a large language model as their information processing unit for core reasoning, planning, and decision-making. They can also utilize bespoke tools and recall past actions and conversations to generate contextual awareness. Whereas in the past, specialist MAS agents required rigorously defined rules and frameworks to operate, LLM agents begin their operation having already learned from vast amounts of open data. A generic knowledge base enables LLM agents to adapt to complex environments in real-time (Li et al., 2023). This new paradigm makes agents much more flexible and scalable. LLM agents can interact with humans in natural language, coordinate or compete with other LLM agents, use custom tools to perform tasks such as search and optimisation, query documents and databases, and search the web.

Large multi-national corporations like Walmart and Siemens are already experimenting with agentic LLMs to automate their supply chain tasks. Supply chain information system providers, such as SAP, Microsoft, and Google, are actively developing supply chain agent platforms and interoperability initiatives to enable cross-organizational supply chain automation.

Academic agentic supply chain research is in its infancy, with research fragmented across computer science, operations research, and manufacturing engineering. Early research points to its potential for overcoming supply chain inefficiencies through rapid access to data, improved planning (Menache et al 2025), and cross-organisational negotiation (Janelli et al 2025), but also warns against agentic AI mirroring human bias (Chen et al 2025), challenges in the verification of output and lack of precise language leading to wrong decisions (Menache et al 2025). As this field continues to evolve, it is crucial for researchers to stay informed about the latest developments, raise awareness of potential challenges, and contribute to the growing body of knowledge on the application of agentic AI in supply chain management.

We thus propose to organise a special issue for the International Journal of Production Research on agentic AI in supply chain management in this new era. This Special Issue (SI) seeks to integrate multi-disciplinary research from various perspectives on shaping agentic automation in the supply chain. We welcome a variety of submissions and perspectives. Technical solutions, advanced modeling, mixed-methods, rigorous quantitative and qualitative empirical research, experimental and analytical methodologies with practical industry, managerial, and policy implications are welcome. The following potential topics are relevant, but are not exhaustive, and we encourage creativity in research and investigations:

• Discourse on utilising agentic systems for complex scenarios in supply chain management: Risk and disruption management, logistics and supply chain optimisation, transportation routes, inventory management, quality control, demand forecasting, and warehouse planning and location, and cash flow predictions
• Supply network design with agentic technology: Supplier relationship configurations, agentic digital twins to simulate inventory flows, sustainability implications across the supply chain, circular supply chains, supply chain visibility, and supply chain financing
• Interorganisational agentic systems: Effective multi-agent negotiation and coordination, the design of mediative and persuasive agentic systems, preservation of organisational privacy during multi-agent communication
• Hybrid systems: Integration of agentic systems with blockchain, IoT, Omniverse, and traditional multi-agent systems
• Emergence and Complexity: Unintended consequences of agentic deployment at the system scale, governance, trustworthiness and safety, centralised versus decentralised control, human-in-the-loop agentic systems
• Technical challenges: Performance evaluation, efficient task division, ablation analysis and back testing, sensitivity analysis, agentic architectures operating in high uncertainty environments, long-term horizon reasoning, overcoming hallucinations

References

Bubeck, Sébastien, Varun Chandrasekaran, Ronen Eldan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Peter Lee, et al. Sparks of artificial general intelligence: Early experiments with GPT-4. arXiv preprint arXiv:2303.12712 (2023).

Chen, Y., Kirshner, S. N., Ovchinnikov, A., Andiappan, M., & Jenkin, T. (2025). A manager and an AI walk into a bar: does ChatGPT make biased decisions like we do?. Manufacturing & Service Operations Management. DOI: 10.1287/msom.2023.0279.

Jannelli, V., Schoepf, S., Bickel, M., Netland, T. and Brintrup, A., 2024. Agentic LLMs in the Supply Chain: Towards Autonomous Multi-Agent Consensus-Seeking. arXiv preprint arXiv:2411.10184.

Menache I, Pathuri J, Simchi-Levi D, Linton T (2025) How generative AI improves supply chain management. Harvard Business Review 104(1–2):86–95.

Li J., Zhao Z., Yang C., Huang S., Lee L. -H. and Huang, G. Q. ChatSync: Large Language Model Enabled Spatial-Temporal Knowledge Reasoning for Production Logistics Synchronization, IEEE Internet of Things Journal, doi: 10.1109/JIOT.2025.3603073.

Xu, L., Mak, S. and Brintrup, A., 2021. Will bots take over the supply chain? Revisiting agent-based supply chain automation. International Journal of Production Economics, 241, p.108279.