From prediction to action: AI rewrites supply chains

As geopolitical disruptions multiply, agentic AI could help supply chains anticipate disruptions and act faster

The disruption in the Strait of Hormuz is a stark reminder of how quickly geopolitics can become a supply-chain crisis. Shipping traffic through the strategic waterway has fallen dramatically from normal levels, with energy supplies, freight costs, shipping schedules and sourcing decisions all affected. For supply-chain managers, the challenge is no longer simply to know what has happened, but to anticipate what could happen next, and act before disruption cascades through the network.

This is where the next generation of artificial intelligence could change supply-chain management. The evolution is moving from traditional analytics to predictive AI, then generative AI and now agentic AI: systems that can reason towards a goal, plan a sequence of actions and execute tasks with limited human intervention. The distinction is important. Traditional AI might flag a likely shortage; an AI agent could analyse inventory, supplier availability and logistics options, recommend a response and, within pre-authorised boundaries, initiate the required action.

Connecting predictions with operational decisions

Consider what this could mean during a disruption such as Hormuz. An AI agent could continuously monitor geopolitical developments, shipping movements, supplier information and inventory levels; identify which shipments or plants are at risk; evaluate alternative routes or suppliers; calculate cost and service implications; and present the supply-chain manager with an actionable response. The manager remains accountable, but the machine compresses hours of data gathering and analysis into minutes.

Amazon offers a glimpse of this direction. Its supply-chain operations already use AI for demand forecasting, delivery-location accuracy and intelligent robotics. In 2026, Amazon also launched Amazon Supply Chain Services, making its logistics network and AI forecasting capabilities available to other businesses. Its forecasting models help optimise inventory placement by predicting what customers will want, where and when. The significance is not simply automation; it is the growing ability to connect prediction with operational decisions.

Agentic AI in Supply Chain systems

An especially instructive example comes from Singapore, where AWS and A*STAR's Advanced Remanufacturing and Technology Centre co-developed a Logistics AI Agent. The agent can aggregate information from ERP, transportation, warehouse and customer systems, provide natural-language answers and support actions such as shipment updates and purchase-order interventions. AWS says the system can eliminate up to 50 per cent of manual lookup and reconciliation work. This illustrates a critical advantage of agentic AI: instead of forcing a manager to move between multiple systems, the agent can reason across fragmented information and turn it into an operational response.

Global logistics companies are also moving in this direction. DHL has deployed agentic AI for shipment booking and data enrichment, automating more than 70 manual data fields and reporting a 40per cent productivity gain in one application. It has also introduced AI-powered item identification that uses a photograph of a shipment to generate a customs-compliant description, improving data quality and reducing clearance delays.

The importance of resilience

India is not a bystander. Delhivery has introduced an AI-agent-powered autonomous transport management system to automate freight procurement, shipment planning, execution and invoice reconciliation. It has also launched an AI-driven customer-support agent. These developments suggest that agentic AI is moving beyond experimentation into core logistics processes in India.

Yet the emergence of agentic AI should not be confused with fully autonomous supply chains. AI is only as good as the data, processes and systems surrounding it. Poor master data, fragmented IT systems and weak process discipline cannot be solved simply by adding an AI layer. More importantly, not every decision should be delegated to a machine.

That shift matters because resilience increasingly depends on speed. In a volatile environment, a company that detects a disruption ten hours earlier can secure capacity, reroute cargo or activate an alternative supplier before competitors do. Agentic AI therefore has the potential to turn resilience from a periodic contingency exercise into a continuous capability, provided humans establish clear rules for what an agent may decide, what it must recommend, and what always requires approval.

The Manager's role is changing

The question is no longer whether AI belongs in the supply chain, but how much autonomy should it be given.

The supply chain manager of the future may therefore have a very different role. Instead of spending disproportionate time collecting information, reconciling spreadsheets and chasing routine updates, managers could supervise a network of AI agents. Their value would lie in setting objectives and boundaries, challenging machine recommendations, managing exceptions, negotiating with stakeholders and making strategic decisions when the consequences are too significant for autonomous action.

The real transformation, therefore, is not humans versus AI. It is the emergence of supply chains in which humans and AI continuously sense, predict, decide and act together. The next competitive advantage may not be better prediction, but faster intelligent action.

The author is an alumnus of IIM Ahmedabad and Professor of Practice at IILM University, Gurugram with an interest in AI, Technology and Strategy; Views presented are personal.