The aggregate quality spreadsheet, connected to the ERP

The problem

Two parallel worlds, and one person copying cells between them every month.

Two parallel worlds coexist: the real spreadsheet, where the live aggregate data lives, and the official ERP, updated weeks later by someone spending hours a month copying and pasting. That lag makes it hard to cross-reference each recycled-aggregate batch with its CE-marking traceability.

• The live data is in the spreadsheet: grading, fines below the sieve, moisture and the count of wood, plastic and gypsum the technician runs on each batch of recycled aggregate produced by the plant.
• The official ERP has a transcribed version of it, incomplete and weeks out of date, kept alive by somebody spending hours a month on copy and paste between two screens.
• That lag is what breaks the cross-check between a batch and its CE-marking traceability, which is precisely what a customer reusing the aggregate — or the site engineer accepting it — asks to see.
• And the transcription errors surface at the worst possible moment: when a delivery is questioned and the sheet has to prove what that batch actually was, not what somebody typed it as.

How it fits the IRIS system

Connect in listen-inward mode — it watches the shared folder, never the person.

Connect's "listen inward" mode: it watches the shared folder (not the person) and reads the delta when the file is saved. A grounded LLM parses it, shows the data for the technician's validation and, once confirmed, pushes it to the ERP via API.

Nobody is asked to change the sheet, and that is deliberate: asking the quality technician to standardize a file they have refined over years is the fastest route to a system nobody uses and a second spreadsheet appearing in a private folder. iLEAN reads the one that exists, in the shape it is in today, including the columns somebody added for the last job and never removed.

See the full IRIS architecture →

Before and after

Today's quality control versus the sheet connected to the ERP

Aspect	Today	With iLEAN Connect
Gap between sheet and ERP	Weeks	Zero
Manual transcription	Hours a month	None
Batch to CE-marking cross-check	Rebuilt by hand, weeks later	Automatic, batch by batch
Copying errors	Surface when questioned	Eliminated
The spreadsheet format	Must be respected	Read as it is, changes included
Quality module licenses	Several	One, the technician who validates

Before: weeks of lag and hours a month spent transcribing. After: zero latency and automatic per-batch traceability cross-referencing.

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.

• 💰 CFO angle: the ERP bills per named user. If several lab and quality people log in purely to enter results, that is a 50-80 % reduction in quality-module licenses, recurring every year.
• Estimated payback 4-9 months, before counting anything the lag itself costs you.
• Hours a month of an expensive quality technician recovered from copying cells between two screens that should have been talking to each other.
• And per-batch traceability that stops being assembled in a hurry the afternoon a customer asks for it, because it is already there.

CFO angle: a typical 50-80% reduction in ERP licenses for the quality/lab module (from several licenses down to one). Estimated payback 4-9 months. Estimate to validate.

And the fair question from the production manager

"That sheet has been passed between three technicians over the years — will it hold?" — that is the anticipated case, not the exception. Parsing does not depend on fixed cell positions: it interprets columns, units, merged cells and margin notes without asking anybody to standardize anything or to stop adding a column when a job needs one. It is an anchored task, where the best models drop below 1.5% error [1], and the technician sees and validates the parsed batch before anything is pushed into the ERP.

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