Zero manual typing between the ERP and the labeler

The problem

One keystroke on the labeler can compromise a whole batch.

one wrong keystroke on vintage or back-label numbering compromises an entire batch in front of the regulatory council — and it only surfaces weeks later.

• The labeler prints batch, vintage and the numbered back-label assigned by the regulatory council. On a busy day it runs several orders, each with its own data. It is a modern machine with an API — and it is isolated from the ERP.
• So at every bottling order someone types on its panel what the ERP already knows: batch, vintage, variety, the back-label range for that order.
• One wrong digit on the vintage or on the back-label numbering is enough to compromise the whole batch in front of the regulatory council. Back-label numbers are controlled one by one; a skipped or repeated range has to be explained.
• And the error does not show on the line: it surfaces weeks later, when the cases are already with the distributor and the only fix is a withdrawal. The cost of that withdrawal is far above the cost of the bottling run itself.

How it fits the IRIS system

Connect stitches two modern systems through their APIs, replacing neither.

Connect stitches two modern systems together via API. When the ERP flags the start of a bottling order, it automatically publishes batch, vintage, variety and numbering to the labeler.

Nothing on the line changes for the operator except that the fields arrive already filled in. The error that used to depend on a tired finger at the end of a long bottling run simply stops being possible, because nobody types the vintage any more. And the operator gains the minutes that used to go into copying fields at every order and format change.

See the full IRIS architecture →

Before and after

The isolated labeler versus the labeler fed by the ERP

Aspect	Today	With iLEAN Connect
Vintage and batch on the label	Typed at every order	Published by the ERP
Back-label numbering	Entered by hand	Assigned from the active order
Human-origin labeling errors	Several a year	Zero
When an error shows up	Weeks later, at the distributor	It cannot happen at the source
Non-conformity risk with the council	Permanent	Drastically reduced
Labeler and ERP	—	Both kept, neither replaced

several typing errors on vintage/batch per year → zero human-origin labeling error.

Impact estimate

Impact estimate per bottling order — 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.

• Fast payback, estimated 3-6 months.
• Several vintage and batch typing errors a year drop to zero human-origin labeling errors.
• Drastic reduction in the risk of a non-conformity before the regulatory council, and of a withdrawal from the distributor. That risk is the real reason this case pays back so fast.
• And the operator stops copying by hand what the ERP already has, at every order and every format change. With several formats bottled in the same week, that adds up quickly.

fast payback, 3-6 months; drastic reduction in the risk of non-conformity before the regulatory council. Estimate to validate.

And the fair question from the production manager

"Wouldn't a full integration project with the labeler vendor be cleaner?" — it would, the day it reaches the top of the list, and that day rarely comes. This stitching uses the API the labeler already has and needs no shutdown window. Mapping ERP fields to the labeler's is an anchored task, where the best models drop below 1.5% error [1], and the first orders are checked side by side before going live.

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