Shortage triage finds the components that will stop a build in the coming weeks, prices each one by the orders it puts at risk, and resolves it with an expedite, an approved alternate, a transfer from another site or a re-sequence, while supply that arrives too early is pushed out.
Take the Monday chapter 1 opened on, and add one role built from the blueprint the way the PO cancellation agent was. At 07:40 L. Tan, supply planner, opens the shortage queue. Overnight Juno has read the planning run at plant 1210, the supplier portal and the stock at every location, and lists 14 components that will stop a build in the next ten working days, ranked by revenue at risk. Her 1,240 messages are still there; the 14 are already found. They are the same 14 shortages the A1 scene of chapter 3 reads on this Monday: there the role is designed to rank them for L. Tan to act on; here it is designed to draft the transfer and the expedite as well.
First is a capacitor for controller board 7741-X: 3,200 pieces short on 3 October, because Acme Components GmbH confirmed a week late, with two Northbay Logistics orders pegged to it. Juno proposes two moves. Transfer 1,800 pieces from location L04542, which holds eleven weeks of cover, and expedite the remaining 1,400 with Acme at a quoted premium of $640.
Under version 3 of the shortage policy the transfer sits inside L. Tan’s limit, so she accepts it, and the stock transport order number comes back from SAP S/4HANA at 07:52. The expedite adds cost, so it waits for D. Okafor, who approves it at 08:15. The purchase order change goes to Acme through the supplier portal, and the acknowledgement is stored against the rule with both names.
Planners spend more than half their time sorting dashboards and filters, and the first job vendors give their agents is late supply: finding the alternative source and executing it with the planner’s approval (Kinaxis, 2 April 2025) 16. A top-10 pharmaceutical company cut the steps to find one inventory risk from 40 clicks to 4 and reports up to ten times planner productivity, a vendor-published figure (Kinaxis, 17 October 2025; Supply Chain Digital, 28 October 2025) 17 22.
Every large suite now names an agent for it. Oracle lists an item shortages assistant, a planning stockout advisor and a planning advisor for exceptions, and its inventory planning application is built to resolve stockouts faster (Oracle, product page and 29 June 2026) 32 31. Blue Yonder’s inventory operations agent recommends alternate sourcing or expediting, automated with human approval (Blue Yonder, 5 May 2025) 33.
SAP’s production planning and operations agent checks material, capacity and scheduling before release and recommends workarounds, generally available in the first quarter of 2026, with a planning assistant for exception management due in the fourth quarter (SAP, October 2025 and May 2026) 23 24 26.
The cost of doing it by hand is measured. Expedite spend runs at 3% of logistics cost for top performers and 10% for the rest, and 49% of expedite events trace to inaccurate forecasts (APQC, via Logility, 30 April 2026, a single source) 63. Xpress Boats cut expedited shipping cost by 50% with agents on its ERP (Infor, 20 April 2026) 56. Widest vendor coverage in the chain and the clearest price on the manual path: that is why it ranks first.
At 07:00 the material planner opens the stock and requirements list in the ERP, or the exceptions screen of the planning suite, and filters the messages: reschedule in, reschedule out, stock below safety stock, coverage gaps. The shortage list is built in a spreadsheet: part, plant, short quantity, date, and the production and customer orders pegged to each.
Then the calls start. The buyer checks the portal or e-mails the supplier for a new date. The planner checks approved alternates in the bill of materials, stock at other plants and hubs, and asks the scheduler whether the order can move. A daily shortage meeting walks the top 20 items and picks a fix for each.
Each decision is keyed into the ERP by hand: date changes, stock transport orders, production order reschedules, sales order re-confirmations. Half a day to two days pass from signal to executed fix for a cross-site item, an estimate, and finding one inventory risk took 40 clicks at one large pharmaceutical company before agents (Kinaxis, 17 October 2025).
The manual path from the note, and the designed path from the scene. Showcase estate; the designed role is yet to be built.
Shortage triage would be built from the same parts as the PO cancellation agent, and it reads much of the same data: the planning run, the purchase order lines and the supplier confirmations at plant 1210 are already registered and mapped in the showcase estate. What changes is the objective, the actions and who approves what.
| Row | Kind | What it tests or does | Data it reads | Then |
|---|---|---|---|---|
| 01 | Check | The component is short inside ten working days | planning run, pegging, commit dates | carry on |
| 02 | Check | A donor location keeps six weeks of cover after the move | stock by location, demand | row 03 · else row 04 |
| 03 | Action | Create the stock transport order | quantity, transfer value | under $10,000: L. Tan accepts |
| 04 | Action | Draft the expedite with the quoted premium | schedule line, freight quote | to D. Okafor with the reason |
| 05 | Route | A re-sequence or a hot lot | production order, routing | to the scheduler |
A plausible policy sentence for the role, and the planning table it would compile to. Illustrative values; the mechanics are the product’s.
The market runs this use case at draft-and-approve: the vendors cited above execute the alternative with the planner’s approval, and the Gartner note of 20 May 2026 points early autonomy at high-volume actions with a low cost of error. The blueprint designs the acting roles, demand-supply gap resolution and shortage escalation to sourcing, at A2, and the monitoring roles, the planning run review and the coverage check, at A3. The two agree, and a client would start where the PO cancellation agent started.
| Element | Setting | Set by | Changes |
|---|---|---|---|
| Objective | Keep the next ten working days clear to build at plant 1210 at the lowest added cost | L. Tan, planner | New version |
| Data scope | Planning run, orders and pegging, purchase order lines, confirmations, stock at every location, alternates and freight quotes; plants 1210 and 1220 | A. Novak, data lead | New version |
| Rules | The shortage policy and its planning table, version 3 | L. Tan with the Planner Agent | New version, simulated first |
| Trigger | After the nightly planning run, on a moved receipt date or a pull-in, and at 07:00 before the shortage meeting | L. Tan, planner | At once |
| Actions permitted | Create a stock transport order · draft an expedite · propose an alternate · propose a re-sequence · push out early supply | L. Tan, owner | New version |
| Limits | Transfer value under $10,000 with six weeks of cover left at the donor · premium freight under $2,500 a line · 40 actions a day · confidence floor 0.80 | L. Tan and J. Harrison, finance | New version |
| Escalation | Expedites and supplier date changes to D. Okafor · alternates outside the list to the component engineer · two customers for one supply to J. Ridley · a four-hour clock | L. Tan, owner | At once |
| Level | A2 Supervised with L. Tan and D. Okafor as the named approvers | J. Ridley, director | Up on evidence, down in one step |
| Feedback | Every accept, edit and reject is recorded and used in the next run | Always on | Fixed |
Every row is a setting or a versioned rule; the level and the limits are the ones a manager signs.
Lima’s agent blueprint is a design of 960 agent roles across five end-to-end chains. It is the map of what Juno can be configured to do, and the first agents are built from it one at a time, the way the PO cancellation agent was.
Additional capability in Plan-to-Produce lists the rest of the map for this chain.
| Measure |
|---|
| Clear-to-build share for the next ten working days |
| Shortage-driven line stops per week |
| Expedite and premium freight spend as a share of logistics cost (baseline 3% for top performers and 10% for bottom performers, APQC via Logility) |
| Clicks and minutes to find one inventory risk (baseline 40 clicks, 4 with agents, Kinaxis) |
| Hours from signal to executed fix |
| Share of shortage actions executed within rules, approved as drafts, edited or rejected |
| On-time delivery and revenue at risk cleared |
Baseline clear-to-build share for the next ten working days first, because it is the objective in one number and the planning run already computes its inputs each night. Hours from signal to executed fix comes next, read from the audit trail, so the 35 minutes of the scene can be set beside the half day to two days of the manual path. Expedite spend as a share of logistics cost is the finance measure, and the APQC benchmark gives it a range to sit in.
The agent stops at the allocation and routes it to J. Ridley with both orders, the revenue at risk on each and the commit dates. Allocation between customers is a commercial decision, so it stays with a person.
Yes, the way the PO cancellation agent moved. Transfers inside the limits build a record; a passing test, J. Ridley’s approval and a live escalation route would let them execute on the agent’s decision, with L. Tan on the exceptions.
It sends the purchase order change through the supplier portal after D. Okafor approves it, and stores the acknowledgement. The conversation with the supplier stays the buyer’s.
Nearly half of these expedites trace back to a forecast that was wrong. Which forecasts are wrong, and who should be looking at them?
Each cycle the agent would score every item-location, let the stable ones through untouched and put only the exceptions in front of the planner.
Bring two systems and one decision your planners make every week. In the first week we register the data, confirm the mapping with your team, answer your questions in ChatLima and run the first agent at the level you choose.
