The forming machine panel, read without touching it

The problem

The machine that decides the seal leaves no trace of how it was set.

The forming-and-sealing machine is where it's decided whether the seal will hold through packing and transport. It's also the most expensive, hardest-to-replace asset: swapping it means weeks of downtime and a six-figure investment. Today nobody correlates its parameters with sealing-defect scrap because the panel's data dies the moment the screen rotates to the next view.

• The forming-and-sealing machine is where it is decided whether the crimped edge will survive packing, freezing and transport. It is also the plant's most expensive and hardest-to-replace asset.
• It is usually more than a decade old, with a proprietary panel and no network connection: swapping it means weeks of downtime and a six-figure investment.
• Forming speed, sealing temperature and cutting pressure show on the screen and die right there, every time the view rotates or somebody taps another tab. Nothing is stored beyond the machine's own short buffer, and nothing leaves it.
• So nobody can answer the question asked in every quality meeting: does this week's sealing scrap concentrate in one formula, one shift, one dough lot or one specific machine setting?

How it fits the IRIS system

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

Connect photo-on-panel mode: a fixed industrial camera points at the forming machine's panel. OCR plus a grounded LLM reads each value and structures it as a time series per batch, cross-referencing which batch and formula was running at that moment.

It does not integrate with the forming machine's software, opens no port and installs nothing on its controller. On a machine the plant cannot afford to stop, that is the difference between an approvable case and a project maintenance never authorizes.

See the full IRIS architecture →

Before and after

Today's forming machine versus the forming machine being read

Aspect	Today	With iLEAN Connect
Forming and sealing parameters	Visible for an instant on screen	Time series per batch
Sealing scrap	A count at shift end	Correlated with its cause
Root cause by formula or shift	Impossible to argue with data	Analysis in minutes
Fine tuning of the machine	In a veteran operator's head	Queryable history
The machine's software and validation	—	Untouched
Replacement capex	In the investment plan	Deferrable

Zero parameter-to-scrap correlation → root-cause analysis per batch in minutes. Machine replacement postponable indefinitely.

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, with the avoided capex of replacing the forming machine inside the calculation.
• Scrap reduction from the parameter-to-sealing-defect correlation that is impossible to make today.
• Weeks of downtime for a machine swap that stop being planned for.
• And the fine tuning of speed, temperature and cutting pressure stops living only in the memory of the operator who has run that machine for twelve years.

Estimated payback 5-10 months, avoiding machine-replacement CAPEX. *Estimate to validate*.

And the fair question from the production manager

“Does a camera reading a screen give quality a serious data point?” — here the task is anchored: fixed panel, 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 instead of accepting it, which is what would invalidate the series in front of an auditor.

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