ICT and AOI stitched to the MES, nothing replaced

The problem

The tester knows the verdict; the system that releases the lot does not.

- The test result (pass/fail, electrical values, serial number) stays isolated in the tester's own software. - To release the lot, someone has to note or review the on-screen result and re-type it into the central system, risking transcription errors.

• The SMT line's ICT or AOI station already produces a result per unit: pass or fail, measured values per test point, and the serial number of the board.
• That result stays isolated in the tester's own software, because there was never time to integrate it properly when the station was commissioned.
• To release the lot, someone has to read or review the on-screen result and re-type it into the central system, with the transcription errors that implies.
• And when a field return comes back, linking that ECU or sensor serial to its in-circuit measurements means opening the tester's logs by hand, station by station.

How it fits the IRIS system

Connect as the API bridge between tester and MES — both stay as they are.

Connect stitches two modern systems via API: whatever happens on the ICT/AOI station is automatically reflected in the MES/ERP as soon as it's generated, including the link to the unit's serial number.

The test station keeps its fixtures, programs and software; the MES keeps its logic. What changes is that the result stops depending on someone reading a screen and typing it again, which is the step where a fail can quietly become a pass.

See the full IRIS architecture →

Before and after

The isolated test station today versus the stitched station

Aspect	Today	With iLEAN Connect
Per-unit ICT or AOI verdict	On the tester's screen and local log	In the MES as soon as it is generated
Link between result and serial number	Typed by hand	Carried automatically
Lot release	Waits for someone to review and re-type	Released on the recorded results
Measured values per test point	Kept only in the tester	Available for trend analysis per lot
Field return on one ECU serial	Search through tester logs	One query in the MES
Retest after repair	Unclear which result counts	Every attempt kept, final verdict clear

Manual transcription of the test result → automatic, immediate reflection in the central system.

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 3-6 months.
• Shorter lot-release time, because release no longer waits for someone to review and re-type test results.
• Lower transcription-error risk: a fail typed as a pass is exactly the error that reaches a customer.
• From manual transcription of the test result to automatic, immediate reflection in the central system, linked to each serial number and available to quality the moment the unit leaves the station.

Estimated payback 3-6 months, reducing lot-release time and transcription-error risk. Estimate to be validated.

And the fair question from the production manager

“What if the integration passes a wrong verdict to the MES?” — here nothing is interpreted: the result travels from one API to another as structured data, which is even more anchored than reading a document, where the best models already stay below 1.5% [1] error. And every message is checked for completeness: a result with no serial number or an unknown test program is held and flagged, never written. A person resolves it with the tester log in front of them.

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