Batch startup in a breadstick and crispbread snack factory — the first label is where the shift is at stake.

The problem

The batch startup is where the chain breaks, not steady-state production.

In a factory of breadsticks, crispbreads or grissini — any regional dry bakery running multi-SKU lines — the line in steady state is stable: once it starts, package after package comes out with little incident. The problem is the SKU change moment, happening several times per shift and concentrating three changes at once:

• The label reel — the ingredient declaration, allergens and format change. The previous batch's label should never appear on the new one.
• The date code — the expiry recalculated for that SKU, with whichever retailer rule applies (different weeks for private label or own brand).
• The seasoning recipe — sesame, cumin, whole wheat, salt-free, with tomato. The silo and the doser change; the intermediate cleaning should be recorded before starting.

The production order arrives on paper from the office (sometimes late, sometimes with a change scrawled in pen on top). The operator reads it, mentally crosses it with the first label out of the printer, and decides whether to start or call the shift lead. It is the shift's natural bottleneck: if they decide right, nothing happens; if they decide wrong, the brand can end in a recall for wrong labeling. And they almost always do it alone, by the cabinet light, with the line stopped waiting.

How it fits the IRIS system

Connect in manual mode — the operator's phone digitizes the paper and the package, without touching the line.

The batch startup needs no new screen on the line or a touched PLC. The information that already exists on paper (the office's production order) and the one that already comes out printed (the first label) needs to reach the system at second zero, without rewriting it and without depending on the operator's memory. That is what Connect in manual mode is for.

The operator photographs the paper production order. Connect digitizes it by OCR. They photograph the first label. Connect reads it and crosses it with the expected recipe. If something does not add up, it alerts that very instant. The person signs — the line does not resume on its own.

How Connect manual operates at the batch startup of breadsticks and crispbreads:

• A photo of the paper production order — the operator or the shift lead photographs with their phone or tablet the production order that came down from the office. Connect digitizes the SKU, the expected recipe, the quantity, the customer and the expiry date to apply. In seconds, the order is in the system with nobody having typed it.
• A photo of the first printed label — before releasing the first package, the operator photographs the first label out of the printer. Connect reads by OCR the ingredients, the declared allergens, the batch code and the printed expiry.
• A cross-check with the plant's recipe wherever it lives — Connect crosses the digitized order and the label with the silo's recipe, wherever it is: a modern ERP, a vertical bakery MES or the Excel the R&D manager keeps in a shared folder. Nothing has to be migrated; it connects where the data is.
• An alert on the earpiece — if the current batch's label declares "nut-free" but the previous batch's seasoning carried almond and the intermediate cleaning is not on record as complete, Connect alerts the operator instantly with a short phrase: "careful, this label declares nut-free but the cleaning after the almond batch is not recorded as closed". The person validates or calls the shift lead; the line does not start on its own.
• The startup file saved — the photo of the order, the photo of the first label and the cross-check with the recipe are archived per batch. When the IFS/BRC auditor asks for the February 14 startup file, there is nothing to reconstruct: it is there.

See the full IRIS architecture →

Before and after

Manual startup by eye vs. startup digitized with Connect

Aspect	Classic startup by the cabinet light	With iLEAN Connect manual
The production order	On paper, read by hand, sometimes with pen changes	A phone photo — digitized by OCR in seconds
The batch's first label	Crossed from memory with the order, the operator's call	A phone photo — crossed with recipe and order at second zero
The SKU change on the line	20-25 min with three paper checklists	4-6 min with two photos and an alert if there is a deviation
The seasoning recipe / silo	An R&D Excel that does not always reach packing	Connect reads it wherever it is — ERP, MES or the Excel itself
The incident notice	The operator goes looking for the shift lead	An instant alert on the earpiece or phone — a short phrase
The file for the IFS/BRC auditor	Rebuilding the startup by hand, weeks later	Photo + cross-check archived per batch, automatic

Impact estimate

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

• Breadstick, crispbread or grissini factory: 1-2 packing lines, 6-10 SKUs per shift, reel and recipe changes several times a day.
• Connect manual pilot on the batch startup — only the operator's phone/tablet, no construction work on the line. First value expected within a few weeks.
• Indicative payback between 4 and 9 months, depending on the frequency of documented startup incidents (wrong label, misapplied expiry, crossed seasoning) and the average recall or rework cost at your retailer.
• Expected reduction of startup incidents ≥30% in the first three months, by simple automatic cross-checking at second zero. Estimate to be validated with your data.
• The hard lever is a single recall avoided: product recovered from the shelf, reverse transport, destruction, brand damage. One recall pays for the pilot.

And the fair question from the production manager

"What if the OCR misreads the label and lets a batch through with the wrong declaration?" — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI limits itself to recontextualizing a specific data point from one medium to another (reading the printed label and comparing it with the expected recipe), the best models brought the error below 1.5% ^[1]. And even then, nothing critical is decided alone: Connect holds the startup and the person signs. The three safety rings are there precisely for this.

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