The lab stops transcribing into the ERP

The problem

Two versions of the truth, and the ERP is the wrong one.

That sheet lives in a shared folder the official system never sees. Somebody transcribes it later, badly and late, or simply does not. And the ERP ends up weeks behind the real data. The uncomfortable moment comes when the customer asks for evidence of a specific test. The question then is not where the data is. It is which of the two versions is the right one.

• Every shift the lab fills in a spreadsheet with the results product conformity rests on: foam density and hardness, seam strength, anchorage pull test, flammability, durability. It is the real record of what was tested.
• That sheet lives in a shared folder the official system never sees. Somebody transcribes it later, badly and late, or simply does not. The ERP ends up weeks behind the real data, and the transcription errors surface at the worst moment.
• The uncomfortable moment comes when the customer asks for evidence of a specific anchorage pull test: the question is not where the data is, but which of the two versions is right, and nobody wants to answer that in front of the customer.
• And five people log into the ERP only to key in results, which is five licenses billed every year for typing, in every plant of the group.

How it fits the IRIS system

Connect in inward-listening mode — the folder, not the person.

Connect in inward-listening mode. iLEAN observes the shared folder, the folder, not the person, and reads the delta the moment the file is saved. It parses it with a model grounded on the test schema. It shows the result on screen to the lab manager for validation and, with that sign-off, writes it into the ERP through its programming interface. One master, fed in real time. And the lab technician keeps working exactly as before. And here is the argument the finance director signs. The ERP is billed per active user. If five people in the lab log into the ERP only to key in test results, that is five licenses. With Connect only the manager who validates needs to log in: one license. A typical reduction of fifty to eighty percent in quality and lab module licenses, with recurring annual savings. And in a multi-plant group that same pattern replicates plant by plant, which is what turns this into a boardroom argument rather than a single-factory one.

The argument the finance director signs is the license count: the ERP bills per active user, and with Connect only the lab manager who validates needs to log in. In a multi-plant group the same pattern repeats plant by plant, which turns a lab case into a boardroom one. And the lab technician keeps working exactly as before: they save the file, nothing else.

See the full IRIS architecture →

Before and after

Today's lab versus the connected lab

Aspect	Today	With iLEAN Connect
Gap between spreadsheet and ERP	Weeks	Second zero
Transcription	Half a day a month, by hand	Zero
Evidence of a pull test for the customer	Which version is right?	One master, by part and serial number
Lab technician's routine	—	Unchanged: saves the file
Who validates	Nobody in particular	The lab manager, on screen
Quality module licenses	Five	One

Impact estimate

Impact estimate — to be validated against your actual invoice.

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 4-9 months, plus the recurring license saving, which is what turns this into a defensible figure.
• 💰 CFO angle: 50-80% fewer licenses on the quality and lab module, every year, in every plant of the group, to validate against your actual ERP invoice.
• Half a day a month of transcription gone, and no more doubt about which version is right when the customer asks.
• Tests cross-referenced against part and serial number, which is what the customer's investigation asks for and what the audit dossier is built from.

Estimated payback 4-9 months · 50-80% fewer licenses on the quality module Estimated payback of four to nine months, plus the recurring license saving. An estimate to validate against your actual ERP invoice, which is the number that turns this into a defensible figure.

And the fair question from the production manager

“Several people have been changing that spreadsheet for years — does it hold up?” — that is the anticipated case: parsing is grounded on the test schema, not on fixed cell positions, so it reads columns, units and margin notes without asking the lab to standardize anything. It is an anchored task where the best models drop below 1.5% error [1], and the lab manager validates before the write.

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