Paperless Batch Start for Gourmet Croquette Plants

The problem

The batch is defined by two sheets of paper that only exist on the line.

Every shift starts with a printed work order and the batter recipe sheet for that SKU. With 18 different croquettes in the catalog, matching the right sheet to the right batch is where origin errors start.

• Every run starts with a printed work order and the batter recipe sheet for that SKU: milk, flour, butter, the filling and the format. With 18 croquettes in the catalog, the sheet changes several times a day.
• Matching the right sheet to the right run is where origin errors start, and in a gourmet range that sheet also carries the chef's version of the recipe and the notes that came out of the last tasting panel.
• Between the shift lead signing it and somebody transcribing it, four to eight hours pass — and over a weekend of campaign production, considerably more.
• During that gap the batch exists on a tray and in a freezer, but exists in no queryable system: not for the store's stock, not for a distributor asking which allergens that lot declares. And in a range built on short runs, two or three batches can start before the first one is typed in.

How it fits the IRIS system

Connect in photo mode — the shift lead keeps filling in the same paperwork.

The shift lead photographs the work order and the recipe sheet at start-up. iLEAN Connect reads the structure and extracts SKU, allergens, format and notes, with zero change to how they already work. The batch is recorded in iLEAN's central memory the instant the line starts, with its SKU and allergens already traced, instead of waiting hours for someone to transcribe it.

It is the lightest of the twelve to deploy and the one the floor accepts fastest, because it asks for no new habit: the shift lead adds one photo to what they already did. That is usually why it is the first case signed off.

See the full IRIS architecture →

Before and after

Batch start-up on paper versus batch start-up captured

Aspect	Today, on paper	With iLEAN Connect
When the batch exists	4-8 h later, more over a weekend	Second zero
SKU and format of the run	Read off a sheet on the line	Recorded with the batch
Allergens of the recipe	Implicit in the sheet	Traced from the first minute
The chef's recipe version	Whichever sheet was printed	Tied to the batch it produced
Handwritten note on the sheet	Lost unless somebody types it	Extracted and attached
The shift lead's habit	—	Unchanged: one photo

Work order and recipe sheet on paper, transcribed hours later → Batch logged with SKU and allergens in zero seconds.

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 4-9 months, depending on how many runs a day the line starts.
• Transcription time the production lead stops spending and gets back on the floor.
• From a 4-8 hour lag to zero-second capture, weekends included.
• And every batch reaches the store's stock and a distributor's traceability query with its SKU and its allergens already attached.

Payback 4-9 months · from 4-8h lag to zero-second capture. Estimated payback for this pilot is 4 to 9 months depending on daily batch volume, plus the transcription time it frees up for the production lead. Estimate to validate against real plant data.

And the fair question from the production manager

"What if it reads the wrong quantity off the recipe sheet?" — reading fields from a document with a known structure is an anchored task, where the best models drop below 1.5% error [1]. And there is a second net: the summary lands on the line tablet and the shift lead confirms or corrects it before anything crosses into the system.

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