The press panel that talks

The problem

In a mixed fleet, the blind presses are not the least important ones.

Those legacy presses still mold critical parts, but their process data dies on the panel screen. When a customer complaint comes in about a part from that mold, there's no way to correlate the actual cycle with the defect. Replacing the press costs hundreds of thousands of dollars and weeks of downtime — a hard investment to justify just for the data.

• Several presses on the floor have been molding for more than a decade: proprietary controller, no network port anybody will open, no path to a manufacturing system.
• They are not parked on easy work — they hold molds for parts that ship every week to a demanding customer.
• Their process data lives and dies on the panel screen: the machine that drifts most is the one that reports nothing.
• So when a complaint arrives on a part from that mold, there is no actual cycle to compare against the defect, and the answer to the customer is built on what the setter remembers.

How it fits the IRIS system

Connect in photo-over-panel mode — the fix comes from outside the machine.

Connect photo mode over the panel. Fixed industrial camera. OCR with a grounded language model interprets every frame, structures the process curve by lot, and cross-references it with the active order in the ERP.

Nothing is installed on the controller, no port is opened and no software is revalidated. A fixed industrial camera reads the panel frame by frame, OCR structures the values and they are matched to the order the ERP says is running on that press. On a machine the plant cannot afford to have down, that is the difference between a case that gets approved and a project maintenance vetoes on day one.

See the full IRIS architecture →

Before and after

A press with no data versus a press with its curve stored

Aspect	Today	With iLEAN Connect
Process curve of that press	Dies on the panel screen	Stored lot by lot
A complaint on that mold	No process to compare against	Root cause in minutes
Cycle time actually achieved	What the quote assumed	Measured, per lot and customer
The original controller	Proprietary and off the network	Untouched, nothing installed
Replacing the press	Hundreds of thousands plus weeks down	Deferrable indefinitely
Presses over ten years old	A blind spot in the fleet	The same data as the new ones

zero process-to-complaint correlation → root-cause analysis by lot in minutes. Press replacement indefinitely deferrable.

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 5-10 months, mostly from scrap avoided once process and defect can finally be correlated.
• Machine-replacement capex pushed out of the plan instead of being justified just to get data.
• The oldest presses stop being an argument in the morning meeting and become a number.
• Real cycle time per lot, which is what tells you whether a job was quoted on the right press.

estimated payback 5-10 months, avoided machine-replacement CAPEX, scrap reduction from process-defect correlation. Estimate to validate.

And the fair question from the production manager

"Does a camera reading a panel hold up in front of a customer?" — here the task is anchored: a fixed screen, fields always in the same place and physical ranges known in advance, and on that the best models drop below 1.5% error [1]. Any frame whose value falls outside the possible range for that press is discarded instead of stored, which is what would otherwise wreck the series in an 8D.

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