The lab spreadsheet posts itself into the ERP

The problem

You pay for an ERP and still operate off a spreadsheet.

Two parallel truths coexist in the plant. The lab's, current and living in a spreadsheet; and the ERP's, weeks behind, which is the one production, administration and management actually see. Transcription costs about half a day a month, and even then, when the inspector asks for one batch's analysis, you go looking in the spreadsheet, not in the system. In other words: you pay for an ERP and still operate off a spreadsheet. And there is a cost almost nobody counts: licenses. If four or five people across lab and quality hold a quality module license purely to enter data, that is four or five recurring annual licenses paid for keystrokes.

• Two parallel truths coexist in the plant. The lab's, current and living in a spreadsheet: alcoholic strength, methanol, higher alcohols, aldehydes, furfural, dry extract, batch by batch. And the ERP's, weeks behind, which is the one production, administration and management actually see.
• Transcription costs about half a day a month, and even then, when the inspector asks for one batch's analysis, you go looking in the spreadsheet, not in the system.
• This is the data the certifying body checks against NOM-006. A batch whose analysis sits in a local file is a batch whose evidence depends on one computer.
• And there is a cost almost nobody counts: if four or five people across lab and quality hold a quality module license purely to enter data, that is four or five recurring annual licenses paid for keystrokes.

How it fits the IRIS system

Connect in listening-inwards mode — the file, not the person.

Connect in listening-inwards mode. The flow:

iLEAN reads the file when it is saved; it does not watch anybody. Every analyst keeps their own column layout and their own naming, the quality manager validates on the tablet, and only then is the analysis written to the ERP through the API. From that point on, one license is enough: the one held by the person who validates. The lab does not change how it works; the ERP finally reflects what the lab knows.

• iLEAN watches the shared folder: it reads the file when it is saved. It does not watch anybody and does not monitor people's activity.
• It parses it with a grounded LLM, tolerant of every analyst using their own column layout and their own naming.
• It ties the result to its batch and leaves it in central memory.
• The quality manager validates on the tablet (case 5).
• Only then is it written to the ERP through the API.

From that point on, only the manager who validates needs a quality module license. The rest of the lab carries on working in their spreadsheet, which is where they are comfortable.

See the full IRIS architecture →

Before and after

Today's lab versus the connected lab

Aspect	Today	With iLEAN Connect
Gap between lab and ERP	Weeks	Minutes
Manual transcription	Half a day a month	Zero
Where the inspector's answer is	In a local spreadsheet	In the system, per batch, signed
Column layout	Must be respected	Tolerated as is, analyst by analyst
Quality module licenses	Four or five	One, the manager who validates
Who is watched	—	Nobody: only the file

From a gap of weeks between real data and ERP data to minutes. From analysis in a local spreadsheet to analysis queryable per batch from the system, with its signature. Half a day a month of transcription freed up.

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.

• 💰 CFO angle: direct saving on ERP quality/lab module licenses, typically −50% to −80% in this sub-sector (from 4-5 licenses to 1), recurring every year.
• Estimated payback 3-7 months: one of the fastest in the catalog, because the saving is contractual and shows up on next year's invoice.
• Half a day a month of transcription freed up in the lab, and the month-end rush to key results before the close disappears with it.
• And the analysis the certifying body checks becomes queryable per batch from the system, with its signature, instead of from somebody's computer.

Direct saving on ERP quality/lab module licenses: typical sub-sector ratio −50% to −80% (from 4-5 licenses to 1), with recurring annual savings in the low thousands depending on the vendor's pricing. Estimated payback 3-7 months: one of the fastest in the catalog because the saving is contractual and shows up on next year's invoice. *Estimate to be validated against the actual ERP invoice.*

And the fair question from the production manager

"Three analysts have been changing that spreadsheet 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 asking anyone to standardize anything. It is an anchored task, where the best models drop below 1.5% error [1], and the quality manager validates on the tablet before anything is written to the ERP.

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