The lab spreadsheet, absorbed in real time

The problem

The lab data that certifications depend on lives in a shared folder.

A shared folder holds a spreadsheet the test lab fills every shift with power draw, leak-tightness and temperature results from sampled units — live data that's critical for product certifications. The ERP still needs someone to transcribe it: half a day a month lost to transcription, and weeks of lag between the real lab data and the ERP data.

• The test lab fills a spreadsheet every shift with power draw, leak tightness and temperature results from sampled coolers. It sits in a shared folder. It has grown over the years: tabs per model, notes in the margins, columns added when a customer asked for a new measurement.
• That data is critical for product certifications and for the energy rating each model declares.
• The ERP still needs someone to transcribe it: half a day a month lost to typing, and seconds of copy-paste where a decimal goes in the wrong column. Most of the people doing it are lab technicians whose real job is testing coolers, not typing.
• Between the real lab result and the ERP there are weeks of lag. When someone asks how a model is performing on power draw, the answer is weeks old. And when a certification body asks for the data behind a model's declared energy figure, the lab and the ERP do not always say the same thing.

How it fits the IRIS system

Connect in inward-listening mode — it reads the saved file, not the person.

Connect "listen inward": iLEAN reads the file when it's saved, without watching the person, parses it with a grounded language model, and stores it in the central memory. After human validation, it pushes the data to the ERP automatically via API.

The lab keeps its spreadsheet, its columns and its way of working. iLEAN reads the file when it is saved, a person validates, and the ERP receives the data without anyone retyping a single cell. The lab's way of working is respected; what disappears is the copy between two worlds.

See the full IRIS architecture →

Before and after

The lab spreadsheet retyped versus the spreadsheet absorbed

Aspect	Today	With iLEAN Connect
Power draw results in the ERP	Weeks behind the lab	Real time after validation
Transcription time	Half a day a month	None
Spreadsheet layout	Whatever the lab uses	Unchanged
Decimal in the wrong column	Possible on every copy	Removed with the copy
Quality/lab module seats	Paid to type results	-50 to -80%
Model energy performance	An old report	A current query

Half a day a month of transcription and weeks of lag → data validated in real time in the ERP.

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.

• 💰 CFO angle: the ERP bills per named user, and several lab people log in only to type results. Estimated reduction of -50 to -80% in quality/lab module seats.
• Estimated payback 4-8 months.
• Half a day a month of transcription and weeks of lag become data validated in real time in the ERP. The lab technicians who typed results go back to testing coolers.
• And the energy performance of each cooler model is visible on current data, not on last month's report. Quality can react to a model drifting on power draw while the lot is still in the plant.

CFO angle: -50 to -80% in quality/lab module ERP seats. Estimated payback of 4-8 months. Estimate to validate.

And the fair question from the production manager

“Our spreadsheet has merged cells, notes and a different tab every month — can it cope?” — parsing a spreadsheet whose columns are known test results is an anchored task, where the best models drop below 1.5% error [1]. And nothing crosses to the ERP until a person in the lab has validated the parsed result, so an odd tab is caught before it lands anywhere.

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