Your cutting line panel has been silent for twenty years. It can speak now

The problem

The parameters that explain cut quality are on a screen nobody records.

The slitter panel shows line speed, recoiler tension, meter counter, knife pressure and shear status. These are the parameters that explain cut quality. And nobody records them. When a claim comes in for edge burr or wavy strip, there is no way to correlate the defect with the conditions under which that particular coil was cut. People argue from memory, and memory does not produce a corrective action: it produces an argument. The classic alternatives — replace the line, or try to integrate the PLC — run into two realities: the CAPEX runs into the millions with months of downtime, and touching machine control means opening the cabinet and negotiating the warranty with the manufacturer.

• Line speed, recoiler tension, knife pressure, meter counter and shear status: the slitter panel shows them all, second by second, and none of them leave the screen.
• When a claim comes in for edge burr or wavy strip, there is no way to correlate the defect with the conditions under which that particular coil was cut. People argue from memory, and memory does not produce a corrective action.
• Replacing the line is a CAPEX in the millions with months of downtime. Integrating the PLC means opening the cabinet and negotiating the warranty with a manufacturer that may no longer exist.
• So the plant keeps a line that works perfectly well and stays blind about it, year after year, while every burr claim gets settled by whoever argues best in the meeting.

How it fits the IRIS system

Connect in photo mode on the panel — solved from outside, not inside.

Connect in photo-on-panel mode. A fixed industrial camera with a stable frame points at the screen. OCR with a grounded LLM reads every frame, structures it as a time series and cross-references it with the active cutting run: which heat and which customer were on the line at that moment. Installation is entirely non-intrusive: nothing is wired into the PLC, no cabinet is opened, no risk is introduced into machine control. The machine stays itself; the only change is that its data now exists outside the screen.

Nothing is wired into the PLC, no cabinet is opened and no port is enabled: the only thing that changes is that the panel's data now exists outside the screen, cross-referenced with which heat and which customer were on the line at that moment. Keeping the old slitter stops meaning going blind, and the decision to replace it can be taken on tonnage and condition rather than on the absence of data.

See the full IRIS architecture →

Before and after

Today's slitter versus the slitter being read

Aspect	Today	With iLEAN Connect
Speed, tension and pressure	Visible, volatile, unrecorded	A time series per run
Correlating a burr claim with conditions	Impossible	Minutes
Which heat and customer were on the line	Reconstructed from paper	Cross-referenced automatically
The machine's control	—	Untouched: no PLC, no cabinet
Manufacturer's warranty	At risk with any integration	Unaffected
Replacing the line	Urgent because it is blind	Deferrable

from zero correlation between cutting conditions and defect, to root cause analysis per run in minutes. Replacing the line stops being urgent because keeping it no longer means going blind.

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.
• Deferral of a line replacement CAPEX in the millions, because keeping the line no longer means going blind.
• Scrap reduction through the correlation between cutting parameters and edge defects that is impossible to make today.
• And a root cause per run in minutes when a claim for burr or wavy strip comes in, instead of an argument: the tension and speed of that coil, at that meter, laid next to the defect.

estimated payback 5-10 months, plus deferral of a line replacement CAPEX in the millions and reduced scrap through correlation between parameters and edge defects. *An estimate to be validated*.

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: fixed screen, fields in known positions and expected physical ranges, where the best models drop below 1.5% error [1]. And the system discards any frame whose value falls outside the possible range for that line — a tension of zero at full speed, a meter counter going backwards — instead of accepting it, which is what would invalidate the series in a claim.

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