Plan-to-Produce turns demand into a supply plan, a schedule and finished goods, and it is the chain where analysts find the most AI already in use.
Plan-to-Produce turns demand into a supply plan, a schedule and finished goods, and it is where analysts find the most AI already in use. The Hackett Group reports 74% of supply chain organisations with AI capabilities in S&OP or integrated business planning and 72% in advanced planning and scheduling (The Hackett Group, 14 May 2026) 8. Gartner sizes supply chain software with agentic AI at under $2 billion in 2025 and $53 billion by 2030 (Gartner, 7 April 2026) 5, yet expects that by 2030 only 5% of organisations implementing planning automation will make 10% of their planning decisions autonomously (Gartner, 19 May 2026) 2.
The money sits in three places. Inventory: retail inventory distortion, out-of-stocks plus overstocks, runs at $1.73 trillion a year, 6.5% of sales (IHL Group, 10 September 2025) 61. Firefighting: planners spend more than half their time sorting dashboards and filters (Kinaxis, 2 April 2025) 16; expedite spend runs at 3% of logistics cost for top performers and 10% for the rest (APQC, via Logility, 30 April 2026, a single source) 63; a demand plan still takes more than five days to build at a typical consumer goods company (McKinsey, 2 March 2022) 12. Downtime: unplanned downtime costs the Fortune Global 500 about $1.4 trillion a year, 11% of revenue (Siemens, via AEMT, 1 August 2024) 62.
The market’s own numbers set the level. IDC, for Kinaxis, finds 6% of supply chain leaders running autonomous decisions at scale today and 41% expecting to within two years, with 52% naming trust in AI decisions as a top barrier (IDC for Kinaxis, 11 August 2026) 14. Capgemini found trust in fully autonomous agents fell from 43% to 27% in a year (Capgemini, 16 July 2025) 15; Gartner placed autonomous planning past the peak of inflated expectations (Gartner, 12 November 2025) 6 and says recommendations with human approval are the norm (Gartner, 20 May 2026) 1. Read together: the market runs these agents at A1 and A2 today, grants A3 for narrow, high-volume, low-cost-of-error steps, and treats A4 as a 2030 ambition.
Fourteen candidates were ranked on adoption evidence, vendor offers, named customer results and measured pain. Shortage triage and clear-to-build merges the two top rows, because the agents on the market detect the mismatch and act on it in one workflow, and it has the widest vendor coverage in the chain 33 32 16. Forecast exception review and parameter tuning are the two activities Gartner names as the place to start, high in volume and low in cost of error, and both carry the most named customer numbers 1. Schedule repair is the Produce half: 72% adoption in advanced planning and scheduling 8, suites shipped for it in 2026 24 30 34, and a production scheduling SmartApp already in production use production · anonymised; its customer evidence is the thinnest of the four, and its page says so. S&OP orchestration scored high on adoption and low on what the agents do; excess diagnosis, rebalancing and allocation fold into the four, or follow below.
Find the components that will stop a build in the coming weeks, price each shortage by the orders it puts at risk, and resolve 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.
Read the use case →Each planning cycle, find the item-locations where the statistical forecast or a planner override is off, explain the cause, let stable items pass through untouched, and route only the exceptions to the demand planner.
Read the use case →Keep safety stock, reorder points, lot sizes, coverage targets and planned lead times current against the service target and the measured variability of demand and supply, and write the approved values into the ERP and the planning suite each cycle.
Read the use case →When a machine stops, a lot is held, yield drops or a component slips, re-check the schedule against material and capacity, propose the re-sequence that protects committed dates, and release only the orders whose material, capacity and routing are ready.
Read the use case →Each page works the use case through the six stages, names who does what at which level, and lists the settings you would make.
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.
Everything below is designed and yet to be built. Each role is configured from the same parts as the use cases above, and a role that matters to you can be the next agent built.
| Measure | What it shows |
|---|---|
| Clear-to-build share, next ten working days | Whether the plan can be built; the shortage agent’s objective in one number |
| Forecast accuracy and bias by item-location and lag | Whether the demand plan is right, and whether overrides help or harm it |
| Share of item-locations touchless each cycle | How much of the plan runs delegated, and how much still needs a planner |
| Service level and days of inventory by segment | Whether the parameters hold the target at the lowest inventory value |
| Expedite and premium freight as a share of logistics cost | The price of firefighting: 3% at top performers against 10% (APQC, via Logility) |
| Minutes from disruption signal to approved reschedule | How fast the floor recovers; hours by hand |
| Actions executed within rules, approved, edited or rejected | The record that moves an agent up the ladder, per agent and per 1,000 decisions |
The plan is right, the parameters are current and the floor keeps its dates. At month end all of it becomes numbers that finance has to close. Does the same machinery serve the ledger?
Reconciliations, journal anomalies, accruals and variance commentary: the four use cases finance is deploying now, with the evidence.
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.