Skin packing of Iberian ham slices — the micro-leak you cannot see becomes a return from the shelf.

The problem

The skin pack sells bright red color — and the micro-leak dulls it without warning.

Acorn-fed Iberian ham sells, to a large extent, on color: a ruby red marbled with white fat that says “this is alive, this is worth what it costs”. Skin packaging was chosen precisely because, by wrapping the slice tightly against the tray, it keeps that color for weeks on the chilled shelf. But the whole system rests on the seal being perfect.

The failure stays invisible while the pack is still in the plant:

1. A folded slice that slips under the film and leaves a micro-channel in the seal.
2. A micro-particle on the sealing jaw (fat, fiber, residue) that opens a path a few microns wide.
3. A film reel change with a different thickness or barrier additive, without readjusting the thermoforming cycle.
4. A mold temperature that drifts during the shift change without anyone noticing until the next destructive test.

Packs leave palletizing backed by a random destructive test every x units. The ones that slip through between tests reach the shelf, oxidize, turn brown and come back as a pallet return — with reverse transport cost, product destruction and brand damage with the retailer. The classic system works 99% of the time. The remaining 1% is a whole pallet coming back.

How it fits the IRIS system

iLEAN Edge does not add one more machine — it seals the crack your thermoformer left exposed.

The problem with skin packing is not that the thermoformer runs badly: it runs exactly as it should. The problem is that the quality signature of the pack is split across four systems (the film specification, the mold parameters, the batch sequence and the destructive test) that almost never talk to each other at second zero. iLEAN acts as the putty that fills that gap, without asking you to replace Multivac, Sealpac or your slicer.

Edge sees and measures the sealing area in line. Connect reads the film specification and the sealing-jaw history wherever they live. The agent cross-references them and, if something does not add up, holds the pack. The person signs — never the other way round.

The iLEAN pieces applied to skin packing of Iberian slices:

• Edge — a terminal with machine vision (CNN) and a differential pressure sensor over the sealing area and the vacuum chamber. It detects a folded slice, fat on the lip, a pressure drop below the historical threshold, and fires the rejector or the stack light in milliseconds, before palletizing. It works without a network — if the plant loses WiFi, Edge keeps holding packs back, because what is critical cannot depend on WiFi.
• Connect — captures the film specification (supplier, batch, thickness, barrier) whether it comes from the ERP, from the MES or from a spreadsheet in the R&D department. And it captures what comes from outside: an email from the film supplier about an additive change, a WhatsApp from the retailer with a new specification, a note from quality control about a sealing jaw that vibrated differently.
• Agent — cross-references the image of the seal, the pressure curve, the film specification, the ham batch and the sealing-jaw history. If there is a deviation, it does not send an email at 10 pm: it holds the pack there and then and alerts the quality manager through whichever channel they use. And it stores the evidence (photo + curve + reason) in the batch dossier.

See the full IRIS architecture →

Before and after

Random destructive testing vs. pack by pack with iLEAN Edge

Aspect	Line with random destructive testing	With iLEAN Edge in line
Control coverage	1 in every x packs, by sampling	All of them, at second zero
Detection of a folded slice under the film	Only if it lands in the test	Before sealing, by AI vision
Pressure drop below the historical threshold	Invisible until the return	Differential sensor fires in line
Response to a film reel change	Manual readjustment at shift change	Connect flags it and the agent tightens control
Evidence for an IFS/BRC audit	Rebuilt by hand from the destructive test log	Per-pack dossier with photo and curve
Operation without a network	n/a	Edge keeps holding packs with its own light

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Slicing + skin thermoforming line for Iberian ham, multi-SKU (100% acorn-fed, field-fed, shoulder), with film reel changes several times per shift.
• Edge pilot on one line (camera + differential pressure sensor + actuator + integration with the film specification via Connect). First value expected within a few weeks.
• Expected reduction in returns for skin pack color/oxidation above 30% against the baseline measured during the immersion.
• Indicative payback between 4 and 9 months, depending on the historical frequency of returns and the cost of a recall at your main retailer.
• The hard lever is a single pallet return avoided — that pays for the pilot.

And the quality manager's reasonable doubt

“What if the AI gets it wrong and lets a pack with a micro-leak through?” — hallucination is a problem of free generation, not of anchored tasks. Here the AI merely recontextualizes data from one system into another (it looks at the sealing area and compares it against a threshold, it reads the film specification and verifies the cycle). In tasks anchored to the source, the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN holds the pack and the person signs. The three safety rings exist precisely for this.

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