Skip to content
DGOS

Solution · Manufacturing

ADD - catch defects before they ship.

Edge AI / AIoT vision on the line. Surface and assembly faults with proof.

Quality and process engineers who need AI defect detection without a forever vision-vendor project.

Request ADD Book consulting

The challenge

Defects leave the line. Proof does not.

Quality finds escapes late - after packing, after ship, or when the customer returns a photo.

  • 01

    Manual inspection cannot keep the takt

    Operators miss surface marks and assembly gaps when the line is fast and tired.

  • 02

    Rejects without a reliable reason code

    Rework piles grow, but nobody can show which station and which defect type drove them.

  • 03

    Customer claims without line-side evidence

    Disputes turn into opinion. There is no clip tied to the SKU and timestamp.

Sound familiar?

Talk through your challenge See how we solve it

By the numbers

Late defect discovery costs more than the camera.

Escapes and disputed rejects burn quality and customer trust.

  • Late find

    many defects still surface at pack or customer return

    *1

  • Opinion

    rework piles without station-tied reason codes

    *2

  • No clip

    claims lack line-side evidence at the right timestamp

    *3

  • *1 Common quality pain in discrete manufacturing - inspection capacity lags takt.
  • *2 Plant quality interviews during DGOS discovery - reject reason codes often incomplete.
  • *3 Customer claim cycles frequently restart without synchronized video proof.

How we solve it

AI defect detection on the line clock.

Edge cameras catch surface and assembly defects where they happen - with reject cues and proof clips.

  1. 1

    Catch defects before pack

    Surface and assembly checks at the station - not after the truck leaves.

  2. 2

    Rejects with station context

    Reason codes and clips tied to SKU, station and time for quality reviews.

  3. 3

    Optional PLC stop / divert

    Handshake when you want the line to act - not only an alarm tile.

This is the simpler path.

Request ADD Or start with consulting

ROI conversation

Catch it at the station.

Savings show up as fewer escapes and faster dispute close.

  • Earlier

    defect catch

    Optical checks where the fault happens - not after ship.

  • Proof

    on every reject

    Clips tied to SKU, station and time.

  • Scoped

    model set

    Start with costly escapes; expand by station.

Targets depend on lighting, camera geometry and how tightly SKUs are scoped.

How it works

From part survey to live ADD

  1. 01

    Pick stations & parts

    Which defects, which SKUs, which camera angles this quarter.

  2. 02

    Train scoped models

    Edge AI for your surface and assembly faults - not every defect forever.

  3. 03

    Wire cues

    Operator alerts, reject lights, optional PLC divert.

  4. 04

    Quality reports

    Clip-backed trends by station and SKU.

Ready to map step one?

Book a discovery call Book consulting

What's in the package

ADD for the line that ships.

Scoped for plants that need optical defect detection without a forever vision project.

  • Line-side cameras with edge AI / AIoT for surface and assembly defects
  • Reject / rework cues tied to station and SKU
  • Clip evidence for quality disputes and audits
  • Optional PLC handshake for stop / divert
  • Scoped model set for your parts - not every defect forever

After go-live

What teams notice first

  • Fewer escapes to customers

  • Faster root-cause with clip proof

  • Quality data that matches the station clock

Related use case: Production & patrol

Want outcomes like these on your site?

Request this package Discuss a custom plan

Stop shipping defects you already filmed.

Show us one station and three defect types. We will sketch the ADD cell before you commit.

Request ADD Or start with consulting

Built on

Computer vision Edge AI Modbus OPC UA

FAQ

AI Defect Detection - FAQ.

Quality and process leads sizing camera-based defect detection on the line.

Is this a full ADD / AOI machine cell?

It is AI defect detection on your line cameras / edge AI - scoped for surface and assembly defects you define. We do not pretend to replace every dedicated AOI vendor for every PCB micron class.

Which defects first?

Usually the top escapes by cost or customer claim - scratches, missing parts, wrong orientation, gap / seal issues - proven on one station before plant roll-out.

Can it stop the line?

Optional. Many plants start with operator cues and reject lights; PLC stop / divert comes after trust in the model.

Lighting and camera placement?

Part of the scope. We survey station geometry, glare and takt so models see what humans would call a defect.

How does this relate to OEE?

OEE is line and people visibility. ADD is quality defect detection. They share cameras and edge when useful - they are not the same package.

Tell us the problem — we'll shape the solution.

A short discussion. No pitch theater. We listen, then compose a clear path.

Schedule a discussion Read perspectives