Episode Summary
In this episode of the Optimistic Outlook In Five, Lauren Espin explores how Procter & Gamble (P&G) is using industrial AI to reduce manufacturing waste and improve production efficiency. Traditional machine vision systems can be difficult to reprogram for products with soft, textured materials and complex packaging. P&G’s Visual Inspection Cockpit (VIC) uses deep learning models running on Siemens Industrial Edge to inspect every product in real time, identify defects at full production speed, and recommend adjustments to reduce scrap.
What makes VIC especially powerful is that plant engineers can configure inspections, train AI models, and update the system themselves, putting expertise directly in the hands of the people who know the production line best. Built as a reusable application on the Siemens Industrial Edge platform, VIC can be commissioned five to ten times faster than traditional custom vision systems. The results are already making an impact, with scrap reduced by 10–20%, depending on the product, demonstrating how scalable industrial AI can improve sustainability, productivity, and manufacturing performance.
Full Transcript
I'm Lauren Espin, and this is The Optimistic Outlook In Five, a quick update about an important development that will change the future.
Every production line in the world makes things that don’t meet the standard of the company and have to be thrown away. It’s what we call scrap.
It might not be a huge percentage in a modern factory, but a line running thousands of units a minute does not need a high failure rate to create a lot of scrap.
A piece of scrap is not just wasted raw material. It is usually a finished product or close to it. It already consumed the material, the water, the energy, the machine time, and a slot on the line that something sellable could have used.
Reducing scrap is good for the planet, and it is good for the bottom line.
The only way to reduce scrap is to fix whatever is causing it. Which turns the whole problem into an inspection problem, essentially: how much of what you make can you actually inspect, while it is moving at high-speed down a production line
This problem is typically solved with machine vision. But any time the packaging, product formulation, or environment changes, complex and time-consuming re-programming is required.
For P&G, the problem was even more complex. Traditional machine vision is challenging enough to re-program for a product line with rigid, identical outputs, but for many of P&G’s products, two units off the same machine, both made correctly, do not look identical.
That’s because P&G makes many products that are soft and textured, made of material that shifts and stretches and wrinkles on its own, and does all of that even more when moving at speed down a line. Then add packaging that is heavily printed and heavily decorated, and you have a recipe for a lot of scrap.
So P&G moved to an industrial AI-based inspection system that can be re-configured much faster, can spot more defects in real-time, and can make recommendations to adjust the production line to reduce the number of issues.
The system is called the Visual Inspection Cockpit, or V.I.C. P&G's deep learning models run on Siemens Industrial Edge, on industrial computers with Nvidia GPUs, installed next to the production equipment rather than in the cloud.
The VIC reads live camera images, inspects every single product that passes, and talks straight to the controller, so it can reject one individual item at full line speed without stopping anything.
The piece that makes it especially compelling is a tool that lets the plant's own engineers configure the inspection, train the model, and update it themselves. This means that instead of another team coming in to re-configure machine vision, the people who already know that line, and already know what a real defect looks like on it, can change the system without filing a ticket and waiting.
There’s another piece to this that’s even more compelling for the industry as a whole.
Siemens did not deliver a custom build for one plant. The whole system is set up as a reusable application on the Siemens Industrial Edge platform, with the infrastructure and integration patterns already in place. A new deployment gets commissioned five to ten times faster than a traditional bespoke vision system. That is the difference between spending a lot of time and money on an impressive pilot, vs. the same investment in a system that is up and running across a global manufacturing footprint.
And it’s working. Depending on the product, scrap is down between ten and twenty percent since implementing VIC.
What changed here is that the expertise needed to run industrial AI moved off a specialist team and onto the floor, in a system that can be rolled out to factories around the globe.
I'm Lauren Espin, and this is The Optimistic Outlook.
Please follow the Optimistic Outlook for the latest updates on industrial technology and manufacturing.
Episode details:
Host: Lauren Espin, Siemens USA Corporate Communications
Published: October 27, 2026
Duration: 5 minutes
Related links:
- Optimistic Outlook home
- Listen on Apple | Spotify
