The efficiency spreadsheet stops living apart from the ERP

The problem

The real numbers live in a spreadsheet the ERP never sees.

two parallel worlds — the real spreadsheet (live data) and the official ERP (data re-typed weeks late by someone who spends hours a week copy-pasting and making mistakes).

• Every shift, the continuous-improvement lead fills in efficiency, minutes produced and waste per sewing line in a shared spreadsheet. That is the live data. It is what the line leads look at, and what gets argued over in the morning meeting.
• The ERP holds the official version, re-typed weeks late by someone who spends hours a week copy-pasting. By the time it is there, the shift it describes is long gone.
• Every copy-paste is a chance to put line 3's efficiency on line 4, or a work shirt order's waste on a coverall order. The error is small on the day and large in the month-end report. Nobody notices until the monthly report disagrees with what the line leads remember.
• Meanwhile the plant pays production and planning module licenses for people whose only interaction with the ERP is that re-typing. It is a cost nobody questions because it has always been there.

How it fits the IRIS system

Connect listening inward — the shared folder, read the moment the file is saved.

Connect "listen inward" watches the shared folder and reads the delta the moment the file is saved; an anchored LLM parses every cell, it's validated on screen, and synced to the ERP via API.

Nobody changes their spreadsheet. Not the template, not the habit, not the shared folder it lives in. Connect reads the delta the moment it is saved, a person validates it on screen, and the ERP receives it via API. The licenses that only existed for re-typing stop being needed. And the efficiency the ERP shows is finally the same one discussed at the line in the morning.

See the full IRIS architecture →

Before and after

The parallel spreadsheet versus the synced spreadsheet

Aspect	Today	With iLEAN Connect
Lag between floor and ERP	Weeks	Seconds after saving
Re-typing	Hours every week	Zero
Line or order swapped by copy-paste	Discovered at month-end	Caught at validation
Production and planning licenses	One per person who types	Only the validator
The improvement lead's spreadsheet	—	Unchanged
Efficiency by line and style	Two versions that disagree	One version, in the ERP

weeks of lag → zero seconds. Hours-per-week of transcription → zero.

Impact estimate

Impact estimate — to be validated against your actual license 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.

• Estimated payback 4-9 months.
• CFO angle: typical 50-80% reduction in production and planning module licenses, because only the validator needs one. It is usually the figure that gets the finance team into the conversation.
• Hours per week of transcription go to zero. Nobody copies a cell from one screen to another again. That time goes back to the improvement work the lead was hired for.
• And the lag between the floor and the ERP goes from weeks to seconds. Planning works with this week's efficiency, not last month's.

CFO angle — typical 50-80% reduction in production/planning module licenses (only the validator needs a license), estimated 4-9 month payback. *Estimate to validate*.

And the fair question from the production manager

“What if it puts a figure in the wrong line or the wrong cell?” — parsing a spreadsheet whose columns and lines you defined is an anchored task, where the best models drop below 1.5% error [1]. And a person validates the change on screen before it is synced, with the edited cells highlighted, so a misplaced value is seen before the ERP sees it. Formulas and totals are checked too, so a broken sum is flagged instead of being synced.

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