Lab data stops being retyped by hand

The problem

Two parallel worlds, and a person copying cells between them.

Two parallel worlds exist. The real one, the lab folder, where the live data sits and where anybody who genuinely needs to know goes to look. And the official one, the ERP, with transcribed, incomplete data lagging by however long the person doing the transcription lags. When a customer complaint arrives, the two have to be reconciled and they almost never match. Half a day a month copying and pasting is the least of it: the serious part is that auditable evidence is being retyped by hand, with the error rate that implies. And there is a hidden operational cost: an out-of-tolerance value is only discovered when somebody gets around to transcribing it, with the press having produced for hours against that condition.

• The lab produces data all day: CMM reports on the instrument panel carrier, color readings against the OEM master plaque, gloss, paint cross-hatch adhesion, resin moisture before drying, melt flow index of each incoming lot.
• All of it lands in a shared folder, which is where anyone who genuinely needs to know goes to look. The ERP holds a transcribed, incomplete version lagging by however long the person transcribing lags.
• When a customer complaint arrives, the two have to be reconciled and they almost never match. Auditable evidence is being retyped by hand, with the error rate that implies.
• And an out-of-tolerance value is only discovered when somebody gets around to transcribing it, with the press having molded for hours against that condition.

How it fits the IRIS system

Connect listening inwards — the folder, not the people.

Connect listening inwards. The system observes the shared folder — not the people — reads the delta when the file is saved, parses it with a grounded model including the PDFs each instrument formats its own way, normalizes against the control plan, shows the result for human sign-off and pushes it into the ERP quality module through its interface. If a value falls out of tolerance it does not wait for transcription: it flags it there and then and alerts the lead.

The alert does not wait for transcription: a color reading outside the OEM tolerance or a moisture value over the drying limit fires while the press is still running, not at the end of the month. The rest of the case is the CFO's: the ERP is licensed per named user, and the people who log in purely to type stop needing to.

See the full IRIS architecture →

Before and after

Today's lab versus the connected lab

Aspect	Today	With iLEAN Connect
Gap between folder and ERP	Days or weeks	Second zero
Transcription	Half a day a month, minimum	Zero
Color reading out of tolerance	Found at transcription	Flagged while the press is running
Link between result and lot	Never cross-referenced	Automatic, lot by lot
Instrument PDF formats	Each one retyped	Parsed as they come
Quality module licenses	Five	One, the person who validates

from a lag of days or weeks to second zero · from half a day a month transcribing to zero · from an out-of-tolerance result found at transcription time to one that fires while the press is running · from quality data never cross-referenced with the lot to automatic lot-by-lot linkage.

Impact estimate

Impact estimate — to be validated against your actual ERP 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, on top of the license saving.
• 💰 CFO angle: the ERP bills per named user. If five people across lab and quality hold a quality module license purely to key in results, that is five licenses; with Connect only the person who validates needs one. Typical reduction of 50-80% on that module, saved every year.
• Half a day a month of copying and pasting, gone, along with the transcription errors in auditable evidence.
• And the out-of-tolerance color or moisture result fires while the press is running instead of weeks later, when the lot has already been molded and shipped.

CFO angle — the ERP is licensed per named user. If five people across lab and quality hold an active quality module license purely to key in test results, that is five licenses; with Connect only the person who validates needs one. Typical reduction of 50-80% on that module's licenses, saved every year. Estimated payback 4-9 months on top of the license saving. *Estimate to be validated against the actual ERP invoice.*

And the fair question from the production manager

"Every instrument formats its report differently — does it hold up?" — that is the anticipated case: parsing does not depend on fixed positions, it interprets the CMM report, the spectrophotometer export and the adhesion sheet as each instrument writes them, and normalizes them against the control plan. It is an anchored task, where the best models drop below 1.5% error [1], and the quality lead signs before anything is pushed.

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