THE DELIVERY NOTE NO LONGER BLOCKS THE TRUCK

The problem

Every traceability chain in the plant starts with typing done in a hurry.

The truck arrives with starch, fat or aromas. The receiving supervisor has the paper delivery note and the labels on each sack or drum. Either the truck waits 45 minutes racking up demurrage while entries are typed line by line into the ERP, or the entry is rushed and a batch ends up poorly traced — something that surfaces weeks later in an allergen alert.

• The truck arrives with starch, fat or aromas, and the receiving supervisor has a paper delivery note plus a label on every sack and drum. The driver is already asking how long this will take.
• Either the truck waits 45 minutes racking up demurrage while lines are typed into the ERP, or the entry is rushed through to get the dock clear. Both options cost money; only one of them shows up on an invoice.
• A rushed entry means a lot wrongly associated with a sack, and that error does not surface at the dock: it surfaces weeks later, inside an allergen alert.
• Which is the worst possible moment to discover that the chain back to the supplier has a hole in it, with a distributor already asking how far the affected product went and how quickly you can tell them.

How it fits the IRIS system

Connect photo-to-analog — nothing at all is asked of the supplier.

Connect photo-to-analog: one photo digitizes the delivery note and another the batch labels (GS1-128, QR, barcode or plain OCR). The system extracts batch, quantity, expiry and supplier; after one validation tap on the tablet, everything enters the ERP.

Two photos replace 45 minutes of typing, and the note is cross-checked against the sack labels before anything is validated. Nobody performs that cross-check today with a driver waiting at the dock and a line asking for material, which is exactly why the errors get in.

See the full IRIS architecture →

Before and after

Receiving as it runs today versus receiving by photo

Aspect	Today	With iLEAN Connect
Time the truck spends at the dock	45 minutes	5 minutes
The lot on each sack or drum	Typed line by line	Read from the label, GS1 or plain
Layout of the note	A different one per supplier	No template needed for any of them
Note against labels	Nobody cross-checks them	Disagreement shown before validating
Tracing back from an allergen alert	Rebuilt with gaps	Complete down to the sack
Active seats in the receiving module	Two or three	Just the validator's

45-minute receiving → 5 minutes. Hidden batch errors → zero. Full traceability by photo and GS1, down to the raw-material sack.

Impact estimate

Impact estimate — to be validated against what the dock costs you.

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 needs two or three active licenses; with Connect only the supervisor validates, without ever opening the ERP — a 50 to 70% reduction, recurring every year.
• Estimated payback 3-8 months, before counting the demurrage that stops being invoiced to you.
• From 45 minutes per truck to 5, on a dock that is usually the plant's first bottleneck of the morning and sets the tone for the whole shift.
• And hidden lot errors go to zero, which is exactly what an allergen alert or a traceability exercise will eventually put to the test.

CFO angle: the ERP's receiving module typically requires 2-3 active licenses; with Connect only the manager validates without opening the ERP — typically 50 to 70% fewer receiving/warehouse licenses. Estimated payback of 3 to 8 months. Estimate to be validated.

And the fair question from the production manager

"Our suppliers' notes are all different — will it cope?" — reading a delivery note is an anchored task even with a different layout per supplier, because the fields you need are always the same handful, and the best models drop under 1.5% error [1]. The second net is the cross-check itself: note against sack labels, with any disagreement raised on the tablet before the supervisor validates the entry. That comparison is something nobody has time to do by hand, so it is new control rather than replaced control.

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