Photo of the production order at batch start-up

The problem

The two sheets of paper that define the batch take hours to exist.

The shift's production order prints from ERP data, and any formula change or quality note gets scribbled on by hand. The first-pack filling sheet is handwritten by quality policy. Both can sit undigitized for hours — when the auditor arrives or the shift changes, someone has to reconstruct from memory which formula and which allergens that batch carried. When several filling formulas run in the same day (in-house brand plus contract manufacturing for third parties), that paper sheet is the only source of truth for the batch's allergens.

• The shift's production order prints from ERP data, but gets written on by hand the moment anything deviates: a filling formula change, a piece-weight adjustment, a dough lot that comes in drier than usual.
• The first-pack filling sheet is handwritten by quality policy, and it is the only source of truth for which allergens that particular batch carried: gluten, egg, dairy.
• Hours can pass between the signing of those two sheets and their digitization, and over a production weekend considerably more.
• When in-house brand and contract manufacturing for third parties run in the same day, that gap is exactly the one that hurts: the batch is formed, cut, crimped and packed, but it exists in no queryable system until somebody finds a quiet moment to type it in.

How it fits the IRIS system

Connect in photo mode — the line lead keeps filling in their paperwork.

Connect photo mode: the line lead photographs the order and the filling sheet with a phone. A grounded LLM extracts formula, batch, allergens, and format, storing them in central memory. No habit change.

It is the simplest of the twelve to deploy and the one that creates least friction on the floor: nobody changes a habit, a photo is simply added to what was already being done. That is usually why it is the first one approved and the first one to prove value.

See the full IRIS architecture →

Before and after

Start-up on paper versus start-up captured

Aspect	Today, on paper	With iLEAN Connect
Digitization latency	Hours, and more over a weekend	Second zero
The batch's allergens	On a handwritten sheet	Queryable by batch instantly
Formula change noted by hand	Lost unless somebody types it	Extracted and tied to the order
Piece weight and dough-to-filling ratio	Scribbled in the margin	History per reference
Answering an auditor	Reconstruct from memory	One query by batch number
The line lead's habit	—	Unchanged: one photo

Hours of latency between start-up and digitization → zero seconds. Scattered paper allergen sheets → instant batch lookup.

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 batches per day and formula changes per shift.
• Hours of latency between start-up and digitization turned into second zero.
• The allergen sheet stops being scattered on paper and becomes a per-batch lookup answered in seconds.
• And the shift handover stops depending on somebody remembering which formula, which piece weight and which dough lot were running when the relief crew walked in.

Estimated payback 4-9 months depending on daily batch volume. *Estimate to validate*.

And the fair question from the production manager

“What if it misreads the filling percentage or an allergen on the 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 goes to the line tablet and the line lead confirms or corrects it before anything crosses into the central system.

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