Delivery note and pallet labels in one photo, dock to ERP

The problem

What is repetitive and high-volume is what fails silently.

Packaging goods-in is repetitive and high volume, which makes it the most predictable source of silent error in the plant. Key it in quickly and preform and cap batches end up wrongly associated. That does not show today: it resurfaces weeks later, when a container defect has to be traced or the licensor asks which concentrate lot went into which production run. Connect in analogue-photo mode recognises each supplier's delivery note layout without asking anyone to change theirs, and extracts the full tabular structure. For pallet labels it reads GS1-128, QR codes, barcodes or plain text by optical recognition. The certificate of analysis is filed and linked to the batch automatically.

• Packaging receiving is repetitive and high-volume, and that makes it the most predictable source of silent error in the plant.
• If it is typed in a hurry, preform and cap lots end up badly linked. And that is not noticed today.
• It reappears weeks later, when a container defect has to be traced or when the licensor asks which concentrate lot went into which production run.
• Meanwhile the demurrage clock is running, pushing in exactly the opposite direction from doing it properly.

How it fits the IRIS system

Connect in photo mode — without asking the supplier to change its delivery note.

One validation tap and it is in the ERP. The summary appears on the tablet, the supervisor confirms, and the data goes straight into the ERP via API: batch, quantity, expiry and supplier. Unloading stops tying up the receiving supervisor for the best part of an hour, and packaging traceability is complete straight from the photo. CFO angle: fewer licences at every goods-in point. The ERP goods-in and warehouse module typically requires two or three active licences per receiving point: receiving supervisor, support and line manager. With Connect only the supervisor validates, and does so without entering the ERP. The typical reduction is fifty to seventy percent of that module's licences, multiplied by every receiving point in the network.

The concentrate certificate of analysis stops being a filed PDF: its values end up tied to the lot and, transitively, to every production run that used it. That is what the licensor asks for when they ask.

See the full IRIS architecture →

Before and after

Today's receiving versus captured receiving

Aspect	Today	With iLEAN Connect
Truck held at the dock	As long as the typing takes	A couple of minutes
Carrier demurrage	Recurring	Eliminated
Preform and cap lots	Badly linked when rushed	Tied at unloading
Concentrate certificate of analysis	Filed	Tied to the lot and its production runs
Delivery note format	One per supplier	Recognized without asking for changes
Receiving module licenses	Several	One, the supervisor who validates

Impact estimate

Estimated impact — to validate 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.

• 💰 CFO angle: 50-70 % reduction in receiving module licenses, recurring every year.
• Estimated payback 3-8 months.
• Carrier demurrage eliminated, truck after truck.
• Avoided cost of a packaging traceability error, which today surfaces weeks late.

On top of the recurring license saving comes the elimination of hauler demurrage and the avoided cost of a packaging traceability failure. Estimated payback between three and eight months. Estimate to be validated against your goods-in volume and your ERP pricing.

And the fair question from the production manager

«Does it read a dirty or creased pallet label?» — most carry standard symbology, which is direct reading; optical recognition only comes in when the label is printed without it. In that case it is an anchored task and the best models drop below 1.5% error [1]. And above all there is the cross-check: if label and delivery note disagree, the discrepancy is highlighted rather than resolved silently.

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