Modern machines still worked by hand: closing the last connection

The problem

Two modern machines, both fed by hand, in the most expensive state of all.

Every keystroke of product and batch at the labeler is a chance to print a hundred wrong labels. And a chemical label does not just carry the batch: it carries the hazard pictograms, the precautionary statements and the product identifier. An error there is not administrative, it is a hazard communication failure with regulatory and safety consequences. On the other side, the real weights per container are the figure that would let you close the batch mass balance — how much went into the tank against how much came out packaged — and today they are lost on a handwritten log or recorded aggregated and rounded. The reason the connection is still open is never technical: it is that closing it is never the emergency of the day.

• The labeler that prints the drum label is modern and has a documented interface. Still, someone keys in product and batch at the start of every run. One wrong finger is a hundred labels with the wrong pictograms, the wrong precautionary statements or the wrong product identifier.
• A chemical label is not administrative: it is hazard communication. A drum that says corrosive when the content is flammable is a regulatory failure and a safety incident waiting in a customer's warehouse.
• The filling scale returns the real weight of every container through its port, and nobody is listening: the weights go on a handwritten log, or into the system aggregated and rounded.
• Without real weight per container the batch mass balance cannot close — liters into the tank against kilos packed — and the bulk shrinkage at month end stays a number without a cause.

How it fits the IRIS system

Connect stitches the two machines through their interfaces, both ways, replacing neither.

Connect as the stitch between two modern systems.

From the ERP to the labeler: product, batch, pack size and the GHS pictograms and precautionary statements that match the product's SDS, published with zero keying. From the scale to the ERP: the real weight of every drum, jug and IBC, aggregated by batch. The machines do not change; their connections do.

• From the ERP to the labeler: product, batch, pack size and the set of pictograms and precautionary statements matching that product's safety data sheet, published through the interface. Zero keying.
• From the filler scale to the ERP: real weight per container, automatically aggregated by batch.
• Neither machine is replaced or reprogrammed: their connections change, not them.
• If the installed equipment does not expose an interface, an adapter is inserted: it raises the initial cost but does not break the case.

See the full IRIS architecture →

Before and after

The isolated machines versus the stitched machines

Aspect	Today	With iLEAN Connect
Product and batch at the labeler	Keyed once per run	Published by the ERP
Pictograms and statements	Chosen on the labeler panel	Taken from the product's SDS record
Weight per container	Handwritten or rounded	Real, per container, in the system
Batch mass balance	Estimated by eye	Closes itself per batch
A wrong label run	Discovered at the branch	Impossible by construction
Either machine	—	Neither replaced nor reprogrammed

from one keying per run at the labeler to zero · from hand-written weights to real weight per container in the system · from a mass balance estimated by eye to a per-batch balance that closes itself.

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 4-10 months, subject to confirming that the installed labeler and scale really expose their interfaces.
• Elimination of the risk of an incorrect hazard label through keying, which is a regulatory exposure rather than a productivity figure.
• A weight database per container and per batch, which is the raw material the shrinkage control of case 12 is built on.
• And a mass balance that closes per batch instead of being estimated by eye at month end.

elimination of the risk of an incorrect hazard label through keying, plus the weight database that enables the shrinkage control of case 12. Estimated payback of 4 to 10 months. *Estimate to validate, subject to confirming that the installed labeler and scale really do expose their interfaces.*

And the fair question from the production manager

“Isn't a classic ERP integration cleaner?” — it is, the day the project reaches the top of the IT list, and that is the problem: this is the classic case of two modern machines still fed by hand because closing the loop was never the emergency of the day. Mapping an ERP product record to the labeler's fields and the SDS pictogram set is an anchored task where the best models drop below 1.5% error [1], and the first label of every run is shown on the tablet for the operator to confirm before the rest print.

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