Process applications (the “where”)

In this process, what can be demonstrated today and what is still theoretical?

This pillar maps where agentic AI can create value across the supply chain: planning, procurement, manufacturing, logistics, inventory, visibility, order promising, and disruption response. For each process, we ask the same questions: What has actually been demonstrated under realistic constraints? What has only been proposed? And what would need to change in the data, governance, or operating model for an idea to become operational?

We avoid turning process names into generic “AI agent use cases.” Instead, each application is examined at the task level. Is the agent retrieving information, reasoning over constraints, recommending a decision, executing an action, or responding to an exception? The level of autonomy, and therefore the required controls, should follow the nature and consequences of the task.

The objective is to help practitioners understand where agentic AI is demonstrably ready, where evidence remains limited, and what conditions must be in place before deployment. Readiness is a property of the process and its operating environment, not simply of the model.