top of page

Defect detection - AI is to ensure the second one is prime.

Using AI to improve defect detection starts with the second one being prime
Using AI to improve defect detection starts with the second one being prime

For two decades, the theme of first-time prime has been branded into my head, demeanor, and strategies to improve quality. If unfamiliar, this is a simple calculation where the numerator is the volume or pieces that went through a production line meeting the specifications. The denominator is the volume of pieces that go through the production line regardless of whether meeting the specifications or not. Increasing this ratio has an obvious, favorable impact on the business. Making first-time prime is an obvious key performance indicator to improve OEE, yield, and the conversion cost of the production line.

How to improve your first-time prime

Staying with the obvious theme, the goal is to make more prime material. If you are making Golden Graham cereal, Flamingo Frose - Bubble Gum Pink Nail Polish, or Goodyear performance tires, the goal is to make fewer defects. With a holistic approach to addressing the D in TIMWOODS, you could change process parameters, shut the line down to make corrections, or choose to run a different order. But how does AI fit into this saying the goal is to make first-time prime? The way it does is to strive to make the second one prime. 

Make the second one prime

To make the second one prime, one must be able to detect the defect the moment after creation. It is not to produce 100 units and then shut down to address the defect. It isn’t to monitor if the defect goes away after adjusting a parameter and sprinkling it with a smidge of fairy dust. It is not to have an inexperienced employee trying to interpret a defect from multiple angles as more defective produce gets produced. The goal is to respond at the moment the defect is created.

Defects are going to occur. If you were to put processes and procedures in place to eliminate all defects, you most likely would not be in business. You would be spending more time preventing than producing. Your anxiety asking you what is going to fail next would have you believe that everything is about to fail. 

Failures will occur. Defects will happen. A millwright will leave a wrench inside a machine at start-up only to scratch the product. A rubber-coated roll will delaminate and create a crease on the produced material. A pigeon will fly into a substation to trip a drive on your mill because your relays were not set precisely right. Again; defects will occur. 

Using AI to improve defect detection

To eliminate defects, the mindset shouldn’t solely be on defect elimination. Sure, continuous improvement and thorough investigations will create corrective actions intended to eliminate defects, but they will still occur. If they did not, your Board wouldn’t need you or your team. Instead, your Board wants you to detect the defect quickly. And to define as soon as possible, it starts by ensuring the second one is prime. 

This is the mindset that is required to improve your first-time prime. If Edward Demming was still alive, I feel confident that this would be a rallying cry of his to promote AI. AI has plenty of hype and an immeasurable potential that will improve our processes and procedures. I challenge you to deploy it from a mindset to make the second one prime. Detecting the defect faster will inevitably improve your first-time prime. 


Recent Posts

See All


bottom of page