Photo of the delivery note, no typing while the cold chain waits

The problem

Traceability starts with typing done under cold-chain pressure.

Receiving perishable raw material trades off traceability against speed: with temperature-sensitive product, every minute typing into the ERP is a minute of compromised cold chain. Rush it, and batches get entered wrong — surfacing weeks later in an untraceable safety alert.

• The truck arrives with meat, poultry, cheese, dough or flour: a delivery note of twelve to twenty lines, one lot per box and labels every supplier prints their own way, some with standard symbology and some with a typed block of text.
• With temperature-sensitive product, every minute typing into the ERP is a minute of compromised cold chain, and the dock is neither a chiller nor a freezer.
• Both exits are bad: either you type slowly and correctly with the product out of the chiller, or you type fast and the lots end up wrongly associated. In practice haste always wins, because the cold does not wait and neither does the driver.
• And that resurfaces weeks later, in the safety alert where you have to prove which filling lot the tray that reached the customer came from. If the box-to-lot association was made wrong on the dock, the chain is broken at the first link and no later traceability repairs it.

How it fits the IRIS system

Connect in photo-to-analog mode — no changes asked of the supplier.

Connect photo-to-analog mode: computer vision reads the delivery note's layout without asking the supplier to change theirs, extracting the tabular structure. For box labels: it reads GS1-128, QR, barcode, or plain OCR. One tap validates on the tablet.

Vision recognizes each supplier's delivery note exactly as it comes, and the box labels by GS1-128, QR, barcode or plain text. Nobody has to negotiate a common format with suppliers, which is the reason this problem has gone unsolved for years in almost every plant in the sub-sector.

See the full IRIS architecture →

Before and after

Today's receiving versus receiving captured

Aspect	Today	With iLEAN Connect
Product out of the chiller	As long as the typing takes	As long as a photo takes
Raw-material lots	Wrongly associated when rushed	Read off the box label
Delivery note format	Different per supplier	Read exactly as it arrives
Note-to-label cross-check	Nobody does it	Discrepancy flagged before validating
Safety alert	Traceability broken backwards	Full chain back to the supplier
Receiving and warehouse licenses	Several	One, for whoever validates

Truck held up while typing in the cold → photo in seconds. Hidden batch errors → zero. Broken traceability → complete.

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.

• 💰 CFO angle: the receiving and warehouse module typically requires several active licenses; with Connect only the supervisor validates: a 50-70% reduction, recurring.
• Estimated payback 3-8 months, depending on trucks per day.
• Cold-chain minutes that stop being spent on the dock, truck after truck.
• And hidden lot errors that stop appearing weeks later, with a safety alert on the table and traceability broken backwards.

Typical 50-70% reduction in receiving/warehouse module licenses. Estimated payback 3-8 months. *Estimate to validate*.

And the fair question from the production manager

“Does it recognize every one of our suppliers' delivery notes?” — yes, and without asking any of them to change theirs: the model extracts the tabular structure with no need for a template per supplier. Labels are read by standard symbology when they carry it and by optical recognition when they do not. It is an anchored task, where the best models drop below 1.5% error [1], and the receiving lead validates before anything gets in.

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