The truck no longer waits while someone types

The problem

Component-to-unit traceability starts with someone typing under pressure at the dock.

The receiving lead loses time typing while the dock backs up. Rush it, and mis-entered lots surface weeks later when a field unit needs to know which breaker lot it came from.

• A truck arrives with breakers, busbars or contactors. The receiving lead has the delivery note and the quality certificate in hand and types them line by line into the ERP, lot by lot, quantity by quantity.
• That takes 45-60 minutes per truck, while the dock backs up, the driver waits and the next delivery queues outside the gate. During an expansion, with more trucks every week, the queue only gets longer.
• Rush it, and a lot number goes in wrong or a certificate is never linked. Nobody notices until weeks later, when a unit in the field needs to know which breaker lot it came from and the record points to the wrong one.
• At that point the traceability of a cabinet energized in a data center rests on a typo nobody can find, and the certificate that proved the lot sits in a folder at the dock.

How it fits the IRIS system

Connect in photo-to-digital mode — the delivery note and the certificate, read as they arrive.

Connect's photo-to-digital mode. Vision AI recognizes the delivery note's layout without asking the supplier to change it, and reads lot labels via QR, barcode or plain OCR. Validation happens with one tap on an industrial tablet.

Suppliers are not asked to change their paperwork. The lead photographs the note and the certificate, iLEAN reads layout, lots and labels, and one tap on the dock tablet sends it all to the ERP. The truck leaves, and the lot arrives in the system already tied to its certificate. Every cabinet built from it inherits that link.

See the full IRIS architecture →

Before and after

Receiving at the dock today versus receiving from a photo

Aspect	Today	With iLEAN Connect
Time to clear a breaker truck	45-60 minutes of typing	Minutes
Quality certificate for the lot	Filed on paper	Read and attached to the lot
Breaker or busbar lot number	Typed by hand under pressure	Read from label by QR, barcode or OCR
Mis-entered lot	Surfaces weeks later	Caught at the dock
ERP seats for breaker receiving	2-3 active licenses	Only the validator, outside the ERP
Lot-to-cabinet traceability	Broken at the first typo	Complete from photo and lot code

45-60 minutes of manual receiving per truck → minutes. Hidden lot errors → zero. Broken component-to-unit traceability → complete via photo and lot code.

Impact estimate

Component receiving: estimated impact, to validate against your dock times and licenses.

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, combining dock time recovered, receiving licenses released and lot errors that no longer need to be chased weeks later.
• From 45-60 minutes of manual receiving per truck to minutes, and hidden lot errors down to zero.
• CFO angle: the receiving module typically needs 2-3 active licenses; with Connect only the lead validates, without logging into the ERP, typically 50-70% fewer receiving and warehouse licenses.
• And component-to-unit traceability is complete from the dock, which is where every later lookup, FAT question and field investigation begins. A supplier lot alert can be answered in one query instead of a week of searching.

*CFO angle*: the ERP's receiving module typically needs 2-3 active licenses; with Connect, only the lead validates without logging into the ERP. Typically 50-70% fewer receiving/warehouse module licenses, estimated payback 3-8 months. *Estimate to validate.*

And the fair question from the production manager

“Every breaker supplier sends a different delivery note — can it really read them all?” — yes, without asking them to change anything. Recognizing a document layout and reading lot codes is an anchored task, where the best models drop below 1.5% error [1]. And when the note and the label disagree, the lead sees the mismatch before confirming, instead of discovering it at kitting.

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