The machines already know what they do; the project needs to find out

The problem

The machine already knows its hours; the project finds out when the mold has shipped.

In a project shop the margin is won or lost in the booking. If machine hours arrive estimated, rounded or late, the real cost of the mold is known once it has shipped — which is to say once nothing can be done with that information. There is a second, less visible and more expensive effect: the next quote's date is promised on the planner's experience rather than on the real availability of the work centers. In a sub-sector where winning or losing a program depends on the date offered, that is margin left on the table.

• Program running, spindle cutting, feed, stop, alarm: that data has existed in the control of the shop's modern machines since day one. It was never stitched to the ERP because there was never time to integrate it properly.
• So what the machine already has exactly gets typed by hand, estimated, rounded or late, and the real cost of the mold is known once it has shipped.
• In a project shop the margin is won or lost in the booking. Cost known at project close is cost nothing can be done about.
• The next quote's date is promised on the planner's experience rather than on the real availability of the work centers, and winning or losing a program depends on the date offered.

How it fits the IRIS system

Connect stitching the control and the ERP through the API — replacing neither.

Connect stitching two modern systems through the API — without replacing the control and without replacing the ERP.

iLEAN reads the stream the control already exposes — MTConnect, FOCAS or the builder's equivalent — ties it to the program and the detail loaded in that machine, and publishes cutting hours, stoppages and alarms with their reason into the ERP. When a stoppage reason has to be assigned, the human gate remains: the machinist picks, the system records. The project lead sees the real cost of the mold accumulating detail by detail while it is still being built.

• iLEAN reads the stream the control already exposes (MTConnect, FOCAS or the builder's equivalent).
• It ties it to the program and the detail currently loaded in that machine.
• It publishes spindle cutting hours, stoppages and alarms with their reason into the ERP.
• When a stoppage reason has to be assigned, the human gate remains: the machinist picks, the system records.

See the full IRIS architecture →

Before and after

Today's booking versus the stitched booking

Aspect	Today	With iLEAN Connect
Machine hours	Estimated and typed twice	Measured and published on their own
Real mold cost	Known at project close	Live, day by day
Stoppage reason	Forgotten by end of shift	Picked by the machinist at the moment
Availability per work center	By feel	A real series
Quoting the next mold's date	On experience	On measured capacity
The control and the ERP	—	Neither replaced

from hours estimated and typed twice → measured and published on their own. From real mold cost known at project close → live cost, day by day. From machine availability "by feel" → a real series per work center.

Impact estimate

Impact estimate — to be validated with your numbers.

The block below is an estimate to be validated against your plant's actual data. We put it forward so the committee has an order of magnitude; we refine it during the assessment.

• Estimated payback 4-9 months.
• The end of double keying: the machinist stops typing at end of shift what the control already measured.
• Real cost per mold while it is still being built, when a detail eating hours can still be resequenced.
• And better reliability on future quotes, which in this sub-sector is pure margin.

estimated payback 4-9 months, between the end of double keying and better reliability on future quotes, which in this sub-sector is pure margin. *Estimate to validate*.

And the fair question from the production manager

"Isn't a classic control-to-ERP integration cleaner?" — it is, the day the project reaches the top of the IT list, and that is the problem: this is the classic case of a modern machine still isolated because there was never time. This stitching replaces neither system, asks for no shutdown window and does not compete in that queue. Mapping the control's stream to the detail loaded is an anchored task, where the best models drop below 1.5% error [1], and the stoppage reason is always a human choice.

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