For manufacturers

Catch the faulty unit, not the complaint.

Anomaly detection and visual QC on every unit, at line speed — trained on your own products, running on your own line.

Why AEY

Catch defects, missing components, and out-of-spec parts before the customer does.

Spot anomalies, not just known faults

It learns what normal looks like and flags anything that deviates — including faults you never wrote a rule for.

Verify assembly and completeness

Every component present, oriented and seated. No incomplete build leaves the line.

Prove quality across every line

One standard across every shift and site, with timestamped evidence per unit.

It knows what normal looks like.

Train on your good units. Anything that deviates — finish, shape, position, colour — is flagged with the frame and a score.

Anomaly detection

Flags the fault you didn't predict.

Rule-based inspection only catches the defects you listed. Anomaly detection catches the rest.

  • Learns “normal” from your own footage — no defect catalogue needed
  • Surfaces novel faults, contamination and drift the day they start
  • Runs on the box at line speed, fully offline
Faulty products

Every unit inspected. No sampling.

The classic visual checks, on 100% of production instead of a clipboard sample.

  • Scratches, dents, cracks and finish flaws at line speed
  • Missing, misaligned or unseated components
  • Timestamped image evidence for every reject

Quality intelligence

Automated quality control, on every line.

Defect Detection

Scratches, dents, cracks and finish flaws, at line speed.

Assembly Verification

Every component present, oriented and seated.

Dimensional Checks

Catch dimensional drift before it becomes a batch of rejects.

Label & Print Verification

Missing, smeared or mismatched markings, caught before dispatch.

Count & Completeness

Short shipments caught on the line, not by the customer.

Yield & Throughput Analytics

Defect rate, first-pass yield and throughput, by line and shift.

Prove quality on every unit

Train a model on your own products and defects in an afternoon — no data-science team.