The metrology spreadsheet reaches the ERP without anyone typing it

The problem

Two parallel worlds with a person copying cells between them.

Two parallel worlds coexist: the real spreadsheet, where the lab works, and the official ERP, where the rest of the company looks. Between them sits a person copying cells. The practical lag runs from a week to a month, and the transcription errors surface at the worst possible moment: preparing a product approval or answering a customer complaint. Half a day a month of an expensive technician, at minimum, goes into copy and paste. The silent risk is discrepancy: the number the lab holds and the number the customer sees not matching. In an audit, that is not something you explain with an apology.

• The spreadsheet holds the live data: dimensions from the measuring machine, pull-off forces from the specimens, burst, impulse and permeation tests, the process capability history the customer asks for at every review.
• The official ERP has a transcribed, incomplete and out-of-date version. The practical gap runs from a week to a month.
• Transcription errors show up at exactly the worst moment: preparing a part approval or answering a customer complaint.
• Half a day a month of an expensive technician, at minimum, goes into copy and paste.

How it fits the IRIS system

Connect in inward listening mode — the file, not the person.

Connect in listening-inward mode. It watches the shared folder — the file, not the person — and works when the file is saved. Step 1 — Reading the delta. When the sheet is saved, iLEAN reads what changed. The spreadsheet does not have to move or change format. Step 2 — Grounded parsing. A grounded model interprets each sheet against its known schema: which column is which characteristic, which unit, which specification. Step 3 — Computation. Process capability, trend, out-of-specification: whatever needs computing is computed on the spot. Step 4 — Human validation and write-back. The quality manager reviews on screen and validates. With their approval, iLEAN writes into the ERP through the interface. The lab never enters the ERP.

The process capability the customer asks for at each review stops being rebuilt: it is in the system, up to date, on the day they ask. That turns an afternoon's work into a query.

See the full IRIS architecture →

Before and after

Today's laboratory versus the connected laboratory

Aspect	Today	With iLEAN Connect
Gap between real data and ERP	From a week to a month	Second zero
Manual transcription	Half a day a month, minimum	Zero
Process capability for a review	Rebuilt	Ready
Copying errors	Show up at the worst moment	Eliminated
Sheet format	Has to be respected	Tolerated as is, even as it changes
Quality module licenses	Six	One, the supervisor who validates

Impact estimate

Estimated impact — to validate 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.

• 💰 CFO angle: the ERP bills per named user. If six laboratory and metrology people log in purely to enter results, that is six licenses: 50-80 % reduction, recurring every year.
• Half a day a month of an expensive technician recovered.
• From a week-to-a-month gap down to second zero.
• And process capability stops being rebuilt for every customer review.

CFO angle — direct ERP license saving. The ERP bills per named user. If six people from lab and metrology log in purely to enter results, that is six quality-module seats; with Connect, only the manager who validates needs a seat. Typical 50-80% reduction in quality/lab module seats, a recurring annual saving that depends on no project at all. Estimated payback of 4 to 9 months, plus that recurring saving. *Estimate to validate* against the ERP invoice and the real named-user count.

And the fair question from the production manager

«Several people have been changing that sheet for years — does it hold up?» — that is the anticipated case: parsing does not depend on fixed positions, it interprets columns, units, merged cells and margin notes without requiring anyone to standardize anything. It is an anchored task, where the best models drop below 1.5% error [1], and the supervisor validates before the push.

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