For Electronics
Error-proof the setup. Catch the defect at the station, not at final test.
High-mix, low-volume assembly. Each product variant requires different setup, calibration, and quality checks. Setup errors cascade into scrap. Quality relies on manual inspection, with all its human fatigue and inconsistency. wiseDo error-proofs at loading, and builds station inspection for your boards on an annotation-to-training pipeline we own end to end.
Book an assessment
The Problem
Precision manufacturing at scale. Manual quality inspection at speed.
With this many SKUs, you're changing setup every two to four hours. Load the wrong recipe or calibrate to the wrong variant and the next 500 units are scrap. And inspection is still manual: someone eyeballing parts, sampling a few measurements, making rework calls under time pressure.
Defects are logged but not classified. Root cause stays buried. The same defect repeats on the next high-mix changeover because there's no systematic learning.
The wiseDo Solution
AI error-proofing at loading. AI inspection at placement.
Recipe management with error-proofing at loading: the system cross-checks the loaded variant against the job order before production starts. Station-level visual inspection is built for your boards rather than shipped as a catalogue model — your defect classes are annotated, trained and optimised on our own pipeline, then sealed and admitted through the same release gates as every other model. Defect taxonomy, severity and cause codes are agreed in the assessment; recipe error-proofing and OEE evidence are available first, inspection follows the model build.
Defects are caught at the station rather than at final test. Rework decisions are logged with timestamps and causes. Every operator correction becomes a reviewed candidate for the next model release — never a silent change to the model running on the line.
Recipe check
Setup verified against the work order before the first part is made.
Defect capture
Classified and logged when it happens, not discovered at final test.
Traceable
Evidence tied to the station, recipe, operator and time.
How wiseDo is deployed here
Six weeks, step by step.
An SMT or box-build line, typically 8 to 20 stations. High mix, fast changeovers, tight recipe discipline.
- 1
Line and recipe walk
We map the line from feeder loading to final test, list the recipes and changeover points, and agree the numbers: first-pass yield, rework rate, escapes.
- 2
Cameras at load, placement and test
Cameras cover feeder loading and manual assembly; a camera is added over any station that lacks one. Test-bench and AOI results are read directly. Edge box installed inside your network.
- 3
Recipe error-proofing calibrated
Sight learns what a correct reel load and correct assembly step look like for each recipe. Wrong reel, missing component or out-of-sequence step is flagged before the board moves on.
- 4
Serial-level genealogy live
Every board carries its lot, reel and operator history from the first station. Test data is attached automatically, so a defect is traceable to a reel and a time window, not a shift.
- 5
Defect classification and holds
Defects are classified as they are found, non-conformance workflows open automatically, and suspect lots are held before they ship.
- 6
Review and scale
At week six: yield, rework and escapes against baseline. Second line onboards with the same recipe library.
What wiseDo brings
The platform components doing the work.
One platform, one data model. Each component below reads and writes the same event spine, so nothing is re-typed and nothing is reconciled at month-end.
Sight
Recipe error-proofing at load and step-level assembly verification, on the cameras you own.
PlantOps
Serial and lot genealogy, test-data capture, NCR workflows, audit-ready quality evidence.
Scheduler
Changeover-aware sequencing that reduces the number of recipe switches per shift.
Connectivity
AOI, test benches and your ERP kept in sync, so quality data lives in one record.
What changes on your floor
A wrong reel is caught in seconds instead of at end-of-line test. A customer return is traced to a reel and a ten-minute window instead of a week. Inspection stops being a person eyeballing parts.