The laser marker stitched to the ERP: no typing, no obsolete revision

The problem

The same data typed twice, and the changes are constant.

The same data gets typed in two places at every changeover, and changeovers are constant because the mix is high. It is the classic case of a modern system that stays isolated because there was never time to integrate it properly. One wrong finger on that panel means a whole batch marked with the previous revision's drawing number: scrap or part-by-part rework, and a nonconformity served on a plate if the batch already shipped. And there is a less visible gap: the MES has no record of what was actually marked, only of what was supposed to be marked. Duplicate typing and missing confirmation are the classic entry route for identification error in this sub-sector.

• The marker engraves the DataMatrix that all the unit traceability the customer requires depends on. It is modern, capable and has a documented interface — and it sits outside the real flow.
• Every part number or drawing revision change forces somebody to pick the right template by hand on its panel, with the same data already in the ERP.
• The changes are constant because the mix is high, so the exposure is permanent.
• One wrong finger on that panel means a whole lot marked with the previous revision's drawing number: scrap or piece-by-piece rework, and a nonconformity served on a plate if the lot has already shipped.

How it fits the IRIS system

Connect stitches the two systems through APIs, replacing neither.

Connect as an integration layer between two modern systems. Neither the marker nor the ERP is replaced: they are stitched together. Step 1 — Publication. When the order is released, the ERP publishes the data set: template, part number, drawing revision, batch, cavity, date, plant code. Step 2 — Push through the interface. Connect pushes it to the marker through its API or command interface. Nobody selects anything by hand. Step 3 — Confirmation back. The marker confirms the load and Connect records in the MES what was effectively marked, not what was intended. Step 4 — Closing the ring. That record is what later allows what was marked to be cross-referenced with the active order and with the plant's other verification rings.

What was engraved comes back confirmed to the manufacturing system: not what was asked to be engraved, but what the machine says it engraved. That distinction is what turns marking into evidence instead of an assumption.

See the full IRIS architecture →

Before and after

The isolated marker versus the stitched marker

Aspect	Today	With iLEAN Connect
Typing the same data	Twice per change	Once, in the ERP
Lot with an obsolete revision	A permanent risk	Eliminated by construction
Changeover time on the marker	10-15 minutes	A tap in the ERP
What was actually engraved	Assumed	Confirmed and returned to the system
With high mix	More changes, more exposure	More changes, the same exposure: none
Either system	—	Neither is replaced

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.

• Estimated payback 3-7 months counting only the changeover time recovered — which is minutes on the critical path of the shift.
• The real value is the error that disappears: a lot marked with an obsolete revision is scrap or piece-by-piece rework.
• And a nonconformity served on a plate if the lot has already left the plant.
• With high mix, the number of changes stops increasing the exposure.

Estimated payback of 3 to 7 months counting only the changeover time recovered, which are critical-path minutes of the shift. *Estimate to validate*. The real value is the identification error avoided: a mismarked batch caught at the customer is paid for in sorting, expedited freight and reputation.

And the fair question from the production manager

«Isn't a classic integration cleaner?» — it is the day the project reaches the top of the list, and that is the problem: it is the classic case of a modern system still isolated because there was never time. This stitching replaces neither system and asks for no shutdown window, so it does not compete in that queue.

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