The spreadsheet that stops being copied by hand

The problem

Storage is billed from a spreadsheet the ERP only sees days later.

That spreadsheet is the real source of occupancy data, but the official billing system depends on someone manually transcribing it periodically. The administration manager spends hours a week copying and pasting, with all the error risk that carries.

• Billing storage by pallet position or occupancy requires a per-client stock and location spreadsheet, updated every time goods come in or go out of any of the warehouses. With several warehouses and dozens of clients, that spreadsheet changes many times a day.
• That spreadsheet is the real source of occupancy data, the one the warehouse team trusts. The official billing system depends on someone transcribing it periodically.
• The administration manager spends hours a week copying and pasting, with the error risk that carries straight into an invoice and, from there, into a client's complaint. A row pasted one line off is enough to bill a pallet position to the wrong client for a whole month.
• And several people keep an ERP inventory license only to maintain that copy, which is a recurring cost nobody associates with a spreadsheet.

How it fits the IRIS system

Connect watching the shared folder — the file, not the person.

The system watches the shared folder — not the person —, reads the delta when the file is saved, and parses it with an anchored LLM. The manager validates on screen and the data crosses over automatically, with no manual transcription.

Nobody is asked to stop using the spreadsheet; it works and the team knows it. Connect reads the change when the file is saved, the manager validates it on screen, and the ERP stops lagging days behind the warehouse. Storage is invoiced on today's occupancy, and the licenses that only existed to keep a copy alive can be dropped. It is the same pattern a multi-client operator repeats with every client that pays for storage by position: one spreadsheet, one person copying it, one ERP that finds out late.

See the full IRIS architecture →

Before and after

The copied spreadsheet versus the connected spreadsheet

Aspect	Today	With iLEAN Connect
Lag between spreadsheet and ERP	Days	Zero-second latency after validation
Admin time	Hours a week copying and pasting	Minutes validating
Occupancy billed to each client	What was last transcribed	What the spreadsheet says today
What gets read	The whole file, by eye	Only the delta since the last save
ERP inventory licenses	Several, to maintain the copy	Only whoever validates
The spreadsheet itself in the warehouse	—	Unchanged

Before: days of lag between the real spreadsheet and the official ERP, with hours a week spent transcribing. After: the data crosses over automatically after validation, at zero-second latency.

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.

• Estimated payback 4-9 months, depending on how many people log into the ERP today only to maintain the copy.
• -50 to -80% on inventory/billing module licenses: only the person who validates needs one.
• Hours a week of manual transcription recovered by administration, every week of the year.
• And storage invoices built on today's occupancy, not on the last time someone had time to copy it, which removes a frequent source of billing disputes.

Payback 4-9 months · -50 to -80% on inventory/billing module licenses. If several administration staff log into the ERP just to maintain this spreadsheet, with Connect only the person who validates needs a license — the rest no longer do.

And the fair question from the production manager

“What if it reads the wrong column and bills a client for another client's pallets?” — parsing a spreadsheet whose structure is known is an anchored task, where the best models drop below 1.5% error [1]. Only the rows that changed are proposed, each tagged with its client and location, and nothing crosses into the ERP until the administration manager has validated it on screen. Every validated delta is stored with its author and time, so the invoice for any client can be traced back to the exact spreadsheet change behind it.

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