Manufacturing, logistics & supply chains
Agents now plan, buy and move goods, and robots are entering the factory floor.
The lines that matter Cost of goods sold, scrap, inventory days and cost per shipment
Agents are moving from pilots into planning, procurement and logistics, and physical AI is entering factories and warehouses. The gains are real, but so is the risk of agents making spending and scheduling decisions without clear limits.
Margins in manufacturing and logistics are thin, so the discipline of the ledger matters: scrap, inventory days, cost per shipment and service penalties are where results show up.
- Thin margins and volatile input costs
- Inventory and working capital tied up across the network
- Labour shortages in operations and warehousing
- Service penalties for late or incomplete delivery
- Operational data spread across plants and partners
- Planning systems that are hard to integrate
- Agents buying or scheduling without spending caps
- Entry-level planner roles at risk
Where AI can move the numbers
Opportunities grouped by Agent Value Ledger category. They are possibilities to test, not promised results.
Revenue
Throughput, on-time-in-full performance, faster quoting
Operating costs
Scrap and yield, routing and load planning, procurement
Capital
Inventory days, spare parts, asset utilisation
Risk
Quality detection, supplier risk monitoring, safety
The lines this sector is judged on
Manufacturing
- Cost of goods sold
- Scrap, yield, energy and changeovers
- Throughput
- Equipment effectiveness and planning
- Inventory days
- Raw-material and work-in-progress planning
- Warranty cost
- Quality detection before shipment
Logistics and distribution
- Cost per shipment
- Routing, load planning and carrier selection
- Asset utilisation
- Fleet and warehouse scheduling
- Service penalties
- On-time-in-full performance
- Working capital
- Billing accuracy and collections
A quality-inspection agent on a production line
Assumptions: cost of goods sold US$200m a year; scrap at 3% of cost of goods; inspection staff cost US$40 an hour.
| Line | How the agent moves it | Measured by | Working | Per year |
|---|---|---|---|---|
| Cost of goods | Defects caught earlier, before more value is added | Scrap rate | 0.5 points × US$200m | +US$1.00m |
| Operating cost | Less manual inspection time | Inspection hours | 8,000 hours × US$40, counted only if redeployed | +US$0.32m |
| Revenue | More good output from the same line | Throughput | Not counted until the extra capacity is sold | Excluded |
| Risk | Fewer defects reaching customers | Warranty claims | Not counted until 12 months of claims data | Excluded |
| Cost of the agent | Cameras and integration, model usage, quality-engineer oversight, validation | Total cost of ownership | Annual run cost | −US$0.60m |
| Net annual value | ≈ +US$0.7m, before throughput and warranty benefits | |||
Hypothetical illustration of the method. The figures are assumptions, not a client result or a forecast.
- Spending caps and override rights for planning and buying agents
- Safety-critical operations remain human-reserved
- Supplier and product liability considerations for AI-influenced decisions
Enterprise AI Transformation
Start with a working-capital and quality diagnostic, then oversee priority pilots with stage gates.
About this serviceSupply Chain, Consumer and Manufacturing in the Age of Agency
Agents now plan, buy, sell and move goods, so the advantage will go to firms that redesign how they delegate decisions, not to those that only add tools.
Read the brief →Planning agents without limits
Situation. A manufacturer's planning agents reorder materials and reschedule lines automatically, but nobody has set spending caps or agreed who can override them.
Our approach. Delegation policies with caps by value and supplier, a ledger for each agent, and a plan for the planner roles that change.
What good looks like. The agents keep their speed, and every material decision has a limit and an owner.
Scenarios and worked ledgers are hypothetical illustrations. They do not describe specific clients.
Start with an executive conversation
A confidential discussion of your priorities, and where AI could move the numbers in your business.