AIdataNewsGlobal AIApplications Guide: How P&G Uses VIC and Siemens Industrial Edge to Keep Inspection on Pace With Production
How VIC fits into an applied AI workflow
AI projects often stall when they cannot keep pace with an already running process. This P&G and Siemens example focuses on a quality-gate use case that aims to stay synchronized with high-speed manufacturing, so Procter & Gamble can extend the approach across its global manufacturing operations. The Visual Inspection Cockpit, or VIC, is positioned as the inspection layer for consumer goods produced on fast lines, checking items without interrupting flow.
When adaptive inspection matters
Some consumer goods are made from soft, patterned, or fragile materials that can move, stretch, or wrinkle while machines run. Fixed custom vision setups can lose practical fit when materials, packaging designs, or production conditions vary. VIC uses Industrial AI to stay useful across these variations and supports product checks in real time.
Step tutorial: operating path for line-speed quality checks
- Identify the production step where visual defects must be caught immediately, especially where delicate materials make inspection variable.
- Place VIC where it can observe products during ongoing manufacturing and keep quality decisions aligned with line pace.
- Use Industrial AI to absorb changes in materials, packaging, and production conditions instead of relying only on fixed rules.
- Commission new installations on the joint architecture, noting that the approach can be five to ten times faster than traditional custom vision systems.
- Process inspection outputs near equipment with Siemens Industrial Edge so operations can trigger alerts or remove nonconforming items.
- Collect historical quality data, then review it with production teams to spot patterns and guide ongoing improvement.
Technology stack checklist
- Procter & Gamble deep learning models
- Siemens Industrial Edge computing platform
- Industrial PCs powered by Nvidia GPUs
- Supporting AI hardware and software
- Siemens-provided infrastructure, software scaling, and ongoing operational support across production sites
Value signals from the P&G and Siemens deployment
The practical payoff appears in scrap control and rollout speed. Depending on the product, scrap rates have decreased by 10 to 20 percent. New installations can be commissioned five to ten times faster than traditional custom vision systems. The system also supports more consistent quality by reviewing products in real time and can automatically launch responses such as alerts or line removal of defective items.
Comparison: fixed inspection and adaptive inspection
| Workflow condition | Traditional custom vision behavior | VIC behavior in this use case |
|---|---|---|
| Material, packaging, or environment changes | Often requires substantial reconfiguration | Uses Industrial AI to adapt while sustaining inspection performance |
| High-speed consumer goods flow | May be harder to keep aligned when products vary | Performs checks during production and keeps quality decisions tied to line pace |
| New installation rollout | Custom setup can take more time | Commissioning can be five to ten times faster |
| Defect response | Not detailed in this deployment description | Edge integration can start alerts or removal of nonconforming items |
Common misconceptions
- That changing product conditions automatically require a full vision-system rebuild. In traditional custom vision systems, substantial reconfiguration is often needed, but VIC uses Industrial AI to adapt to material and design variation.
- That edge processing is only a place to store results. Here, inspection output is handled near production equipment and integrated into operations, enabling immediate actions such as alerts or removal.
- That quality checks are isolated decisions. The system can gather data over time and analyze it, helping teams identify trends for continuous improvement.
Action recommendations for AIApplications readers
- Start where AI has to keep up with speed, not just improve accuracy on offline samples.
- Choose adaptive inspection when materials, packaging, or production conditions are expected to change.
- Measure rollout time as well as defect control; five to ten times commissioning improvement changes project feasibility.
- Confirm that edge integration can connect to operational responses, not only reporting dashboards.
- Use accumulated inspection data for trend analysis and ongoing quality decisions.