The grower's delivery note, from paper to ERP in seconds

The problem

Every trailer waits while someone types, and the errors surface weeks later.

every trailer waits at the scale while the supervisor types the load into the ERP. If they rush, plot or variety gets entered wrong — and that surfaces weeks later as a denomination-of- origin traceability problem.

• The grower's trailer reaches the scale with a paper delivery note: grower, plot, variety, gross weight, tare, net weight and signature.
• The supervisor types it into the ERP while the next trailer waits behind. Tractors with engines running, growers impatient to get back to the vineyard, and a queue that reaches the road. At peak campaign the queue grows faster than the typing.
• Typing in a hurry means a plot or a variety entered wrong, and nobody notices on the day because the grapes are already in the hopper.
• That error comes back weeks later as a denomination-of-origin traceability problem, when the grapes are already wine in a tank and the batch cannot be untangled. By then, nobody remembers which trailer brought the wrong plot.

How it fits the IRIS system

Connect in photo mode towards the ERP — the grower's note as it comes, confirmed with one tap.

Connect photo mode. Vision AI recognizes the delivery note and extracts its tabular structure; for trailer/pallet labels, it reads barcode, QR or plain OCR. Validation with one tap.

The grower does not change their delivery note and the supervisor does not type. The only gesture left is checking a summary, which is exactly the moment where plot and variety errors are caught, before they turn into an origin traceability problem. And because the trailer leaves the scale sooner, the next grower does not lose half a morning waiting in line.

See the full IRIS architecture →

Before and after

Typed intake versus photographed intake

Aspect	Today	With iLEAN Connect
Trailer queue at peak	Growing	Flowing
Plot and variety errors	Hidden until the council asks	Caught on the summary
Grower delivery note format	Typed field by field	Read as it arrives
Trailer and pallet labels	Read by eye	Barcode, QR or plain text recognition
Supervisor's job at the scale	Typing	One confirmation tap
Intake and scale module licenses	Three	One, the supervisor who validates

growing trailer queue at the scale during peak campaign → smooth flow; hidden plot/variety errors → zero.

Impact estimate

Impact estimate at the scale — 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 3-8 months.
• CFO angle: a typical 50-70% reduction in ERP licenses for the intake and scale module, going from three licenses to one. The saving repeats every year the license contract is renewed.
• The trailer queue at the scale flows during peak campaign instead of growing with every load.
• And hidden plot and variety errors drop to zero, before they become an origin traceability issue in front of the regulatory council. Settlements with growers at the end of the campaign start from weights everyone already confirmed at the scale.

CFO angle — typical reduction of 50 to 70% in ERP licenses for the intake/scale module (from 3 licenses to 1), estimated payback of 3-8 months. Estimate to validate.

And the fair question from the production manager

"Every grower brings a different note, some handwritten, some printed by a cooperative." — the model recognizes the table structure of the delivery note without a template per grower, and reads trailer or pallet labels by barcode, QR or plain text. It is an anchored task, where the best models drop below 1.5% error [1], and the net weight is checked against the scale's own reading before the supervisor confirms. A note that does not match the scale is never confirmed by accident.

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