Ten years ago, P&G tracked truck deliveries by calling every driver. Now an AI agent reschedules warehouse and distribution center appointments without human involvement. Milwaukee Tool is implementing Oracle agents for supply planning, shipping, and maintenance. Three supply chain leaders described these applications at an IIoT World panel in December 2025.
Where Do AI Agents Work in Supply Chain Operations?
At P&G, agentic AI handles order and invoice management with suppliers, automates request-for-quote processes, and runs supplier performance assessments across quality, cost, and reliability. Those assessments generate risk profiles that flag when a backup supplier may be needed or when the business should be bid out to other suppliers.
For inventory, agentic AI makes real-time adjustments to safety stock settings based on machine learning. In logistics, P&G replaced manual track-and-trace calls with GPS monitoring and an AI agent that reschedules delivery appointments, freeing teams for higher-value work.
Milwaukee Tool is implementing Oracle out-of-the-box agents across supply planning, shipping, and maintenance. In supply planning, where more than 100 planners review and release recommendations, an agent summarizes the notes entered by multiple planners for a specific item or item-supplier combination.
On the shipping side, an agent automates packing, identifies inventory at the correct locators, and generates shipping documents. In maintenance, Milwaukee Tool is exploring a work order agent so technicians spend less time at computers and more time fixing equipment. Maintenance technicians carry iPads and can call the agent after completing a repair to create the work order, issue raw materials, and report the resource.
How Do AI Agents Optimize Production Lines?
AI agents add a prediction layer to production line control systems that already monitor sensor data and execute control actions in real time. Using live sensor data, agents simulate multiple scenarios and determine the optimal control point for a specific product, run, or shift.
The agent either adjusts the line automatically through the control system or presents operators with the recommended settings and alternative scenarios. The optimization target changes depending on the product, the run, or the shift.
How Do Companies Measure AI ROI?
Measuring AI ROI is difficult because success often means nothing happened and business just ran smoothly. At Milwaukee Tool, the team measures value through avoided cost: fewer expedited shipments, faster recovery, and higher service levels. When the team implemented Oracle four months before the panel, the distribution center’s service level dropped below target, and digital twins helped quantify what would have happened without the system, turning resilience into a measurable business outcome.
CIOs face pressure to adopt AI, machine learning, and agentic AI at the same time, but trying to get involved in as many platforms as possible costs money without solving specific problems. Companies should define their top-priority problem statements first, then evaluate two to three platforms and narrow the scope.
ROI requires a clear before-and-after comparison. The cost of a production stop, or an hour of downtime, has to go into the equation for any AI investment.
The areas with the strongest ROI involve robotics and computer vision combined with AI to automate tasks that are either manual or where manual quality is not good enough. A computer vision system with AI can perform the task better and increase line capacity. But implementation costs are high, technology does not solve every problem, and understanding the problem and its size comes first.
FAQ
1. How does P&G use agentic AI in supply chain operations?
P&G uses agentic AI across sourcing, inventory management, and logistics. In sourcing, agents handle order and invoice management, automate request-for-quote processes, and run supplier performance assessments across quality, cost, and reliability. For inventory, agentic AI makes real-time safety stock adjustments based on machine learning. In logistics, GPS tracking monitors shipments and an AI agent reschedules warehouse and distribution center delivery appointments without human involvement.
2. How do AI agents help maintenance technicians in manufacturing?
At Milwaukee Tool, a maintenance work order agent allows technicians to report completed repairs from iPads instead of returning to a computer. The technician calls the agent, identifies the asset and the repair performed, and the agent creates the work order, issues raw materials, and reports the resource. Technicians spend more time fixing equipment and less time entering data at a workstation.
3. How do AI agents optimize manufacturing production lines?
AI agents simulate multiple scenarios using live sensor data from production lines and determine the optimal control point for a specific product, run, or shift. The agent either adjusts the line automatically through existing control systems or presents operators with the recommended settings and alternative scenarios. The optimization target changes depending on the product being made, the specific run, or the shift schedule.
4. How do manufacturers measure AI return on investment?
Three supply chain leaders at a December 2025 IIoT World panel described different ROI approaches. Milwaukee Tool measures value through avoided cost: fewer expedited shipments, faster recovery, and higher service levels. P&G recommends defining problem statements first, then calculating the cost of an hour of downtime as the ROI baseline. Robotics combined with computer vision delivered the strongest ROI for automating manual tasks and increasing line capacity.
Related from IIoT World
- Why AI Supply Chains Are Moving Away From Data Sharing and Toward Signal Sharing
- 15 AI in Manufacturing Use Cases: Predictive to Agentic AI
- Upleveling Supply Chain Management with IIoT Solutions
Sources:
- IIoT World Manufacturing & Supply Chain Day, December 2025, panel: “Can AI Predict and Prevent the Next Supply Chain Disruption?”
This article is based on a panel discussion, “Can AI Predict and Prevent the Next Supply Chain Disruption?,” with Jaimie McIntyre Horstman, Director of Demand Forecasting at Procter & Gamble; Maria Araujo, formerly VP of Innovation Engineering at McCain Foods; and Srinivasan Narayanan, Oracle Solution Delivery Lead at Milwaukee Tool; moderated by Adam Napolitano of Triple Circle Partners at IIoT World Manufacturing & Supply Chain Day, December 2025. AI tools were used to help summarize and organize the content. Reviewed and edited by the IIoT World editorial team.
Watch the full recording: Can AI Predict and Prevent the Next Supply Chain Disruption?
About the author
