Photo the kit traveler at batch start

The problem

The paper that defines the batch only exists on the floor.

That paper defines which components, which BOM revision and which customer, but today it only exists on the floor until someone transcribes it at the end of the shift. A handwritten engineering change can take four to eight hours to reach the system.

• The kitting traveler says which components, which BOM revision and which customer that work order belongs to. In a multi-customer plant it is the only document tying material to program.
• It gets annotated by hand while the kit is picked: an approved alternate substituted for a reel that did not arrive, a quantity that did not come out whole, an engineering change the customer sent that morning.
• Between the kitter signing it and somebody transcribing it, four to eight hours pass, and during that window the work order exists in no queryable system.
• That is exactly the window in which planning reschedules, purchasing commits reels and SMT asks which program it runs next — all of them working off a paper nobody outside kitting can see. When the customer calls to ask whether their engineering change made it into today's build, the only honest answer is that somebody will have to walk down to kitting and look.

How it fits the IRIS system

Connect in photo mode — the kitter keeps filling in the same traveler.

Connect photo mode. The kitter photographs the traveler with their phone or tablet while building the kit. A grounded LLM extracts the work order, BOM revision, customer, and engineering-change notes, and saves it to central memory. After tablet validation (see the early human verification case), the data crosses into the ERP/MES with no one transcribing anything.

Nothing changes on the floor: the kit is picked the same way and the same paper is signed. What changes is that the work order, its BOM revision and its engineering change exist in the system before the cart reaches the line. It is the easiest of the twelve to deploy and the one that creates least friction, which is usually why it is the first one approved.

See the full IRIS architecture →

Before and after

Today's traveler versus the captured traveler

Aspect	Today, on paper	With iLEAN Connect
Digitization latency	4-8 h	Zero-second
Handwritten engineering change	Reaches the system at end of shift, or never	Captured and cross-checked against the ERP
BOM revision in play	Whatever is on the cart	Declared and checked at kit start
Approved alternate used	A note in the margin	Tied to the work order and the customer
Which customer the kit belongs to	Known only in kitting	Visible to planning and SMT
Kitter's habit	—	Unchanged: one photo

4-8h digitization latency → zero-second. Handwritten, lost engineering change → captured and cross-checked against the ERP instantly.

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 work orders you kit per day.
• From 4-8 hours of digitization latency to zero-second, work order after work order.
• The handwritten engineering change stops getting lost: it is captured and cross-checked against the ERP the moment it is written.
• Transcription time recovered from whoever types travelers today, usually a planner or a production clerk who is doing it instead of planning.

Payback 4-9 months · from 4-8h latency to zero-second. Estimate to be validated.

And the fair question from the production manager

“What if it misreads the BOM revision?” — reading fields from a traveler with a known layout is an anchored task, where the best models drop below 1.5% error [1]. And there is a second net: the summary goes to the line tablet and a supervisor confirms or corrects it before anything crosses into the ERP or MES.

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