The chilling tunnel finally gets logged

The problem

The step that decides whether the carcass can be cut leaves no trace.

The chilling tunnel decides if the carcass is ready for cutting, but its data dies every time the batch changes.

• The chilling tunnel is what takes the carcass down to the temperature the cutting room requires. It is the gate between the rail and the primals.
• Its controller is often fifteen or twenty years old, sits off the network by the manufacturer's own design, and shows temperature and elapsed time on a panel nobody records.
• When the batch changes, the screen resets and the previous curve is gone: what survives is at best a temperature written down once, not the curve.
• So when a cold-chain question lands — from a customer, from the official vet, or from your own incident review — the honest answer today is that the data was never kept. The tunnel almost certainly did its job; what is missing is any way of demonstrating it for the batch being asked about.

How it fits the IRIS system

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

Connect photo mode on the panel. Fixed camera + OCR with a grounded LLM structuring the time curve per batch.

Nothing is installed on the controller, no port is opened and its behavior is not modified. A camera sees the panel and a grounded model structures what is on it, batch by batch. On the machine that governs the cold chain, solving it from outside is the difference between a case quality can approve in a morning and a project that never leaves the meeting room.

See the full IRIS architecture →

Before and after

The tunnel today versus the tunnel being read

Aspect	Today	With iLEAN Connect
The chilling curve	Erased on every batch change	Stored batch by batch
Temperature entering the cutting room	A figure somebody remembers	A logged value with its timestamp
Time inside the tunnel	Estimated from the shift plan	Measured from the panel itself
A cold-chain question	Nothing to show	Curve and batch on screen
The controller	—	Untouched, no new software
Replacing the tunnel controller	Budgeted	Deferrable

zero per-batch logging → digitized curve available for any audit or incident.

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, counting the avoided capex of replacing the controller.
• From zero per-batch record to a curve available for any inspection or incident review.
• Carcasses released into the cutting room on a documented temperature instead of on somebody's judgment about the clock, which is what that decision rests on today.
• And a tunnel drifting slowly — a fan losing performance, a door left open on the night shift — shows up as a trend across batches, while it is still maintenance rather than a batch you have to argue about with a customer.

Estimated payback of 5-10 months, avoided controller-replacement CAPEX. *Estimate to be validated*.

And the fair question from the production manager

“Can a camera reading an old panel really stand as a record?” — the task here is anchored: a fixed screen, fields in known positions and temperatures inside a narrow physical range, where the best models drop below 1.5% error [1]. On top of that, any frame whose reading falls outside the possible range for the tunnel is discarded rather than stored, which is precisely what would invalidate a series in front of an inspector.

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