Cured Iberian lomo, curing and weight loss with AI — every extra point of weight loss is expensive lomo going out as thin air.

The problem

Weight loss is not decided at the end of curing — it is decided hour by hour.

The margin on cured Iberian lomo lives on fine numbers: a high price per kilo, expensive raw material and a weight loss that decides how much sellable product comes out of every batch. And the classic problem is always the same:

1. The cellar's curve is not each piece's curve — the fan leaves one zone drier, one rack sits nearer the door and another at the back, and weight loss comes out uneven within the same batch.
2. The recipe changes and nobody cross-references it with the cellar — a new marinade, paprika from a different lot, a casing caliber that arrived different from the supplier. Data in the ERP/MES/spreadsheet that never reaches the master sausage maker in time.
3. Weight loss shows up when the casing comes off — and by then there is nothing left to do. The batch is taken as it comes: weight above normal on some pieces, weight below acceptable on others.

The master sausage maker knows this — he has spent decades weighing pieces by hand to tell how the curing is going. But he cannot be in six cellars at once, nor cross-reference weight loss with the recipe and the curve by eye. Every percentage point of anomalous weight loss is expensive lomo going out as thin air, and the Iberian trade does not forgive that air.

How it fits the IRIS system

iLEAN is not trying to speed up curing — it is trying to cure to the exact point.

The problem with curing stuffed Iberian lomo is not speed: it is precision that does not scale. iLEAN acts as the putty between the batch recipe, the cellar SCADA and the master sausage maker's knowledge, without asking you to change the drying room or the ERP.

Edge measures the curve and the weight. Connect captures the recipe wherever it lives. The agent cross-references it with the expected curve and calls the master when something does not add up. The person signs — the drying room does not change parameters on its own.

The iLEAN pieces applied to cured Iberian lomo:

• Edge — pH, water activity, temperature and humidity sensors per cellar and per critical point, plus weighing by piece or by rack depending on how deep the deployment goes. It captures the real curve continuously. It works without a network: if the plant loses WiFi, Edge keeps measuring and recording, because critical curing cannot depend on connectivity.
• Connect — captures the recipe whether it comes from a modern ERP, from a vertical MES or from the marinating supervisor's spreadsheet. And it captures what arrives from outside (an email from the casing supplier warning of a caliber change, a call from the cutting room) at second zero.
• Agent — cross-references the real curve with the expected curve for the recipe, the placement of the pieces in the cellar and the weight loss of previous batches. If one zone dries faster than it reasonably should, or if weight loss runs away, it proposes three things: check the fan, move the rack to the back, adjust the cellar humidity. The master sausage maker decides; the agent supports.

See the full IRIS architecture →

Before and after

Manual curing vs. governed curing with iLEAN

Aspect	Manual weighing + the master's notebook	With iLEAN Edge + Connect + Agent
Curve per cellar	One sensor per cellar, if there is one at all	Several points per cellar, continuously
Weighing per piece	Manual sampling, periodic	By piece or rack, recorded per batch
Microclimate detection	When odd weight loss appears at unstuffing	There and then, by zone of the cellar
Cross-reference recipe ↔ curve	In someone's head, batch by batch	Automatic, with the history of similar batches
Batch file	Rebuilt by hand	Per-batch dossier with curves, weights and signatures
Operation without a network	n/a	Edge keeps measuring 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.

• Mid-sized plant with 4-8 curing cellars for Iberian lomo, several recipes (classic marinade, spicy, premium), variable cular casing caliber.
• Edge pilot in one cellar (pH/water activity/humidity sensors + weighing + integration with the batch recipe from the ERP/MES). First value expected within a few weeks.
• Indicative payback between 4 and 9 months, depending on the cost per kilo of anomalous weight loss and how often batches have fallen outside the corridor over the past few years.
• Expected floor for improvement: a reduction of ≥ 30% in the spread of weight loss within a batch caused by cellar microclimates.
• The hard lever: expensive Iberian lomo, without losing a kilo to thin air.

And the master sausage maker's reasonable doubt

“What if the AI proposes an adjustment that throws the whole cellar off balance?” — hallucination is a problem of free generation, not of anchored tasks. Cross-referencing the real curve with the expected curve, the recipe and the placement in the cellar is an anchored task. In anchored tasks, the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN proposes, the master 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.