We’re seeing a huge leap in potential when it comes to what AI can accomplish in industrial settings. Not only can it catch errors, it can provide intelligent insights to prevent them in the first place, reduce waste, save costs, and improve processes.  

Recorded live from Automate 2025, we sat down with Alvin Clark, Senior AI Engineer at NVIDIA, to discuss AI agents in manufacturing and how they’re reshaping industrial inspections. We hear about all the different use cases of AI agents, including error detection and understanding the root cause of errors, saving costs, and even supplying training data to patch the gaps in “tribal knowledge” of manufacturing processes.  

You’ll hear real examples of how AI agents have saved costs and reduced errors drastically when monitoring SOPs and how they can provide multimodal maintenance assistance. Alvin also takes us through how NVIDIA’s Metropolis works at helping developers create visual AI agents and why the next few years will see visual AI inspections take off in manufacturing.  

In this episode, find out: 

• Alvin shares his background in the AI space and why he saw potential in AI earlier than most  
• Why 2012 was the real AI boom and how we saw a shift from the algorithm being king to data being king 
• Alvin’s explanation for what an AI agent does in four stages  
• How AI agents are evolving beyond capturing data to providing intelligence in industrial settings 
• How vision inspection can perform not only real time error detection but also real time failure analysis 
• Use cases for AI agents and examples of how Alvin has seen them most successful 
• How AI agents could also address the skills gap and replace the multimodal “tribal knowledge” we’ll lose when people start retiring 
• Alvin breaks down how Metropolis works to help developers build visual AI agents 
• How simulation, training and data transform what AI agents can accomplish 
• Alvin’s perspective on where manufacturers are in their industrial AI journeys 
• The role of systems integrators in leading the AI revolution 
• What the next phase of AI agents will look like   

Enjoying the show? Please leave us a review here. Even one sentence helps. It’s feedback from Manufacturing All-Stars like you that keeps us going! 

Tweetable Quotes: 

• “Metropolis is a combination of models and tools that are used to build what we call vision analytics. So anytime you're looking at video images and you want to extract information from that, these are the tools that can help you do that.” - Alvin Clark 
• “It's not really the ability to catch the error, it's the ability to, as quickly as possible understand the genesis that caused the error.” - Alvin Clark 
•  ”If I had to kind of describe an AI agent, it is a collection of one or more models that can perceive, reason, and potentially plan and then execute.” - Alvin Clark 

Links & mentions: 

• NVIDIA Metropolis, automating physical spaces and infrastructure with interactive visual AI agents and services 
• AlexNet, a convolutional neural network architecture developed for image classification tasks, like identifying a cat 

Make sure to visit http://manufacturinghappyhour.com for detailed show notes and a full list of resources mentioned in this episode. Stay Innovative, Stay Thirsty.