JIT with a non-compatible supplier — between Toyota purity and European reality there is a dynamic buffer.

The problem

The plant runs JIT. The supplier of the critical part does not.

Any European assembly plant that has tried pure Toyota JIT has lived through this. Three or four strategic suppliers are aligned: short lead time, frequent deliveries, stable quality. But there is always one that is not: the cast part made by a single supplier in Asia, the electronic component with a chronic shortage, the subassembly from the small local supplier that delivers when it can. And that single non-JIT supplier forces one of two bad options:

1. Hold a uniform safety stock across every part number so the line never stops. It ties up capital, inflates the warehouse and goes against everything Lean teaches.
2. Keep pure JIT and take the risk. Every so often the line stops because one part ran out, and the operational damage eats months of stock savings.

The operations director knows it. The planner knows it. And the usual choice is the first one (the uniform cushion) because the second one scares the committee more. The result: operationally the plant drifts away from real JIT and towards theatre Lean — the poster on the wall, the warehouse full.

The backdrop does not help: ocean freight, the backbone of any long chain, went from roughly 1,500 USD per container to more than 20,000 at the 2021 peak, spiked again in 2024 with the Red Sea crisis and then fell on overcapacity ^[3]. Volatility breaks classical JIT — but it also breaks the uniform cushion, because a cushion sized for the normal situation gives no protection in the extreme one.

How it fits the IRIS system

iLEAN does not break JIT — it adds a buffer that breathes with each supplier's reality.

JIT does not fail because of the methodology. It fails because the real supplier data (measured lead time, variance, early warning signals) lives in islands and the planner has no time to process it supplier by supplier. iLEAN acts as the filler between the ERP that knows the order, the supplier that knows its own reality and the warehouse that suffers the consequence.

Connect captures the supplier signal (email, messaging app, transcribed call). Brain calculates the dynamic per-supplier buffer every week. The agent alerts the planner when something moves — the person signs.

The three iLEAN pieces applied to JIT with a non-compatible supplier:

• iLEAN Brain — the calculation engine. It combines, per SKU × supplier: real measured lead time, historical variance, line-stoppage cost of that part, cost of storing it. It calculates the target stock level and recalculates it every week with fresh data. It is not a static ERP min/max — it is a dynamic target.
• iLEAN Connect — external capture. The information that gets ahead of a supplier problem almost never arrives through a sensor: it arrives in an email from the sales rep ("there was a fire in the warehouse"), a message from the carrier ("the truck is not leaving today"), a news item from the country of origin. Connect captures the channels — email, messaging apps, transcribed calls — and puts them into the system at second zero, with nobody forwarding anything.
• iLEAN Agent — it coordinates. When fresh data moves the target buffer, the agent alerts the planner, proposes the action (place an early order, redistribute between customers by SLA, contact the alternative supplier) and drafts the reply. The person signs. The system does not buy without a signature.

See the full IRIS architecture →

Before and after

JIT with a uniform cushion vs. JIT with an iLEAN dynamic buffer

Aspect	Classical JIT with safety cushion	With iLEAN Brain + Connect + Agent
Safety stock per SKU	Same percentage across every part number	Sized by the supplier's real risk
Reaction to a supplier delay	Email from the sales rep → meeting the next day	Connect reads it at minute zero, Brain recalculates, agent alerts
Total tied-up stock	Inflated by a uniform worst case	Typical reduction of 30% or more (estimate to be validated)
Line stoppages from stockouts	Low but not zero, with no traceability of cause	Significant reduction, with documented root cause
Visibility for the committee	The planner's monthly spreadsheet	Live dashboard: SKU × supplier × risk
Knowledge across plants	Each plant manages its supplier alone	Cross-plant data on shared suppliers

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the actual data of your supply chain. We lay it out so the committee has an order of magnitude; we refine it during the diagnostic.

• Assembly plant with a JIT customer (automotive tier-2, industrial electronics, appliances), 80-200 active part numbers, 20-60 suppliers, of which 5-15 are not JIT-compatible (long lead time, high variance or both).
• Deployment of iLEAN Brain + Connect + Agent on top of the purchasing module. First value expected within a few weeks: the first buffer recalculation lowers total tied-up stock on the over-sized part numbers and raises it on the genuinely exposed ones.
• Indicative payback between 4 and 9 months, hard levers: reduction of total tied-up stock (capital released) plus reduction of line stoppages from stockouts (operating cost avoided).
• A reasonable total stock reduction to present to the committee: 30% or more. Service reliability improves at the same time. We measure it before and after on your real supply chain.

And the operations director's reasonable doubt

"What if the AI gets it wrong and leaves the line without a critical part?" — hallucination is a problem of free generation, not of anchored tasks. Calculating a buffer from measured lead time, variance and stoppage cost is the anchored task par excellence: on this kind of task the best models brought the error below 1.5% ^[1]. And even so, the critical call is never made alone: the agent proposes the order, the planner signs. The system does not buy without a human signature.

^[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.

^[3] Ocean container freight volatility (Drewry / Freightos Baltic Index, 2019-2026).