The weighbridge, stitched to the ERP with no manual keying

The problem

Two modern systems, a few meters apart, joined by a person typing all day.

When a truck crosses, someone manually copies the weight and plate into the billing system. A mistyped figure becomes a delivery note or waste document with the wrong weight. This creates billing discrepancies and traceability gaps that surface late.

• Every inbound and outbound truck crosses the weighbridge, and its software records the weighing perfectly well. It is modern, it has a documented interface, and it is simply not connected to anything else.
• So somebody reads the weight and the plate off the display and copies them into the billing system, truck after truck, all day, with a queue forming behind the scale while they do it.
• One mistyped digit becomes a delivery note or a waste document carrying the wrong weight — and both of those are documents a customer, a handler or an inspector will later rely on as fact.
• The discrepancy surfaces at month end, reconciling billed tonnage against what the destination handler says it received, and by then nobody can say which truck on which day it came from.

How it fits the IRIS system

Connect stitches the two systems together through their APIs, replacing neither of them.

Connect stitches two modern systems together via API: without replacing the existing weighbridge software, every weighing automatically publishes weight, plate and client/handler into the ERP.

This is the classic case of a modern system left isolated because the integration project never reached the top of anyone's list. The stitching asks for no replacement, no downtime and no new software on the gatehouse machine, so it does not have to win a place in that queue in order to happen this quarter.

See the full IRIS architecture →

Before and after

The isolated weighbridge versus the weighbridge stitched to the ERP

Aspect	Today	With iLEAN Connect
Typing the weight	Once per truck, by hand	Never
Plate and destination handler	Copied by hand from the display	Published with every weighing
A mistyped digit	A delivery note with the wrong weight	Not possible by construction
Month-end reconciliation	Discrepancies with no traceable origin	Every weighing matched one to one
Time at the gate	A queue building behind the scale	The truck weighs and moves on
Either software	—	Neither is replaced

Before: monthly weight/plate keying errors with associated billing discrepancies. After: zero human-origin error in weighing.

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 3-6 months, counting only the billing and traceability discrepancies that stop appearing at month end.
• Zero human-origin error in weighing, which is the real cost: a wrong weight travels into a delivery note and into a waste document.
• Gate time recovered on every truck, inbound and outbound, on a site where a queue at the scale backs straight up into the loader's run.
• And month-end reconciliation stops being an argument about which truck, on which day, the missing tonnage came from.

Estimated payback of 3-6 months from fewer billing and traceability discrepancies. Estimate to validate.

And the fair question from the production manager

"Isn't a proper integration project cleaner?" — it is, on the day it finally reaches the top of the list, and that day is exactly the problem: the weighbridge has had a documented interface for years and is still being keyed in by hand every morning. This stitching replaces neither system, needs no shutdown window and competes for nobody's project slot. Where a value has to be read from a display rather than pulled from the interface, the task is anchored and the best models drop below 1.5% error [1], with the weighing itself as the physical sanity check.

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