The lab spreadsheet flows into the ERP by itself

The problem

The lab knows the fruit's condition today; the ERP learns it weeks later.

Two parallel worlds coexist in the lab: the real spreadsheet, where live acidity, peroxide, moisture and firmness data lives, and the official ERP, which receives it retyped weeks later by someone who spends time every month copy-pasting and sometimes makes mistakes.

• The quality lab lives in a spreadsheet that has grown season after season: dry matter and firmness on incoming fruit, acidity and peroxide value on the oil, moisture, pH and color on pulp and guacamole after HPP.
• The ERP gets a retyped copy weeks later, entered by someone who spends time every month copying cells and sometimes gets one wrong.
• Meanwhile the decisions that depend on those results — release a lot of oil, send fruit to pulp instead of fresh export — are made on the spreadsheet, outside the system of record, and the ERP only finds out after the fruit or the oil has already moved on.
• And several lab people keep an ERP license only to type results, a cost paid again every year for work that adds nothing.

How it fits the IRIS system

Connect listens inward — the shared lab folder, read on every save.

Connect 'listens inward': iLEAN watches the lab's shared folder -not the person-, reads the delta when the file is saved, parses it with a grounded LLM, validates it on screen with the quality manager, and pushes it into the ERP via API.

Dry matter and peroxide value are the two numbers that decide what the fruit and the oil are worth. When they reach the ERP the same minute they are measured, the commercial decision is made on data the system can trace. And when a customer questions a lot, the result is already next to it.

See the full IRIS architecture →

Before and after

The lab spreadsheet cut off versus the spreadsheet feeding the ERP

Aspect	Today	With iLEAN Connect
Delay to the ERP	Weeks	Zero seconds after validation
Monthly retyping	Hours of a technician	None
Releasing a lot of oil	Decided on a spreadsheet	Decided on data in the ERP
Dry matter by orchard	Scattered across tabs	Tied to each orchard lot
Copy errors	Found when a customer asks	Caught at validation
Quality-module licenses	One per person who types	The manager who validates

Weeks of lag → zero seconds. Monthly transcription time → zero.

Impact estimate

Impact estimate — to be validated with your numbers and your 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: if several lab staff log into the ERP only to enter quality data, now only the manager validates — a typical 50-80% reduction in quality and lab module licenses.
• Estimated payback 4-9 months.
• The lag from lab to ERP goes from weeks to zero seconds, so oil and pulp lots are released on results the system already holds.
• And the monthly transcription time goes to zero, along with the copy errors that surface when a customer asks for a certificate.

CFO angle: if several lab staff used to log into the ERP just to enter quality data, now only the manager validates. Typical 50-80% reduction in quality/lab module licenses. Estimated payback 4-9 months. *Estimate to be validated.*

And the fair question from the production manager

“Our lab sheet has tabs for fruit, oil and pulp and changes every season — will it still read it?” — yes: parsing interprets columns, units and notes as they are, with no fixed template, which keeps it an anchored task where the best models drop below 1.5% error [1]. And the quality manager sees each result on screen before anything is pushed to the ERP. A value outside the physical range of the test is stopped there, not in the ERP.

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