The lab spreadsheet and the ERP, finally in sync

The problem

Two versions of every tank: the real one and the one in the ERP.

two parallel worlds — the real spreadsheet with live analytical data, and the official ERP with data transcribed weeks later, with copy errors.

• The enology lab keeps a spreadsheet per week and per tank: alcohol, total acidity, volatile acidity, pH, free and total SO2, residual sugar, malic acid.
• That spreadsheet is the live truth. The ERP holds a version transcribed by hand weeks later, with copy mistakes along the way that nobody spots until they matter.
• When a tank is about to be racked, blended or bottled, the decision is made on the spreadsheet, and the ERP stays behind as a record nobody fully trusts.
• Meanwhile, several people log into the lab module only to retype what was already typed once, and each of them is a license on the invoice. The lab ends up paying for the ERP twice: in licenses and in hours.

How it fits the IRIS system

Connect in listen-inward mode — it reads the lab's shared folder, not the enologist.

Connect "listens inward" — watches the shared folder, not the person. Reads the delta when the file is saved, parses it with a grounded LLM, validates on screen, syncs to the ERP via API.

The lab keeps its spreadsheet, columns and habits exactly as they are. What disappears is the second typing, and with it the weeks of gap between what the tank really is and what the ERP says it is when someone has to decide a blend. It also means the SO2 history of each tank is in one place when an importer asks for it.

See the full IRIS architecture →

Before and after

The lab spreadsheet alone versus the spreadsheet synced

Aspect	Today	With iLEAN Connect
Gap between analysis and ERP	Weeks	Zero latency after validation
Hours a month retyping	Several	None
SO2 and volatile acidity history per tank	Scattered across files	In the ERP, per batch
Transcription slips	Found when blending or bottling	Removed at the source
The lab's spreadsheet at the winery	—	Kept exactly as it is
Lab and quality module licenses	Four	One, the winemaker who validates

weeks of lag → zero latency; hours per month transcribing → zero.

Impact estimate

Impact estimate — to be validated against your ERP vendor's rate card.

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.
• CFO angle: a typical 50-80% reduction in ERP licenses for the lab and quality module, going from four licenses to one, recurring every year. That is the number the finance team usually asks about first, and it repeats at every renewal.
• From weeks of lag to zero latency, and from hours a month spent transcribing analyses to none.
• Figures to validate against your vendor's actual rate card, not a generic price list.

CFO angle — typical reduction of 50 to 80% in ERP licenses for the lab/quality module (from 4 licenses to 1), estimated payback of 4-9 months. Estimate to validate against the actual vendor rate card.

And the fair question from the production manager

"Our spreadsheet has merged cells, notes in the margins and a tab per month — will it really parse it?" — that is the expected situation: parsing does not depend on fixed positions, it interprets columns, units and margin notes as they are. It is an anchored task, where the best models drop below 1.5% error [1], and the winemaker validates on screen before anything is synced to the ERP.

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