In the rapidly evolving world of industrial technology, few people have as deep and diverse a background as Erik Udstuen. A chemical engineer by training, Udstuen cut his teeth in the pulp and paper industry before catching the entrepreneurial bug. Now on his fifth startup, he’s leading the charge at TwinThread to bring the power of scaled analytics and artificial intelligence to process manufacturers.
On a recent episode of the Augmented Ops podcast, Udstuen shared his journey and vision for the future of industrial AI. Early in his career, he implemented neural networks to predict equipment failures – long before the term “AI” was in vogue. Later at GE, he saw firsthand the tremendous benefits of applying advanced analytics and machine learning to large fleets of assets like gas turbines and aircraft engines.
“I walked away from that experience believing that not just the power generation industry should benefit from this scaled approach to analytics,” Udstuen recalled. “We really set out to build TwinThread to address the needs of other process manufacturers, things like consumer products, specialty chemicals, pulp and paper.”
At the heart of TwinThread’s solution is the digital twin. But Udstuen is quick to point out that it’s much more than just a slick 3D visualization. The digital twin serves as a normalizing layer, standardizing data across an enterprise’s diverse equipment, naming conventions, and data sources. This enables the holy grail of building an analytics model once and deploying it many times.
Of course, the ultimate goal is to turn that data into actionable insights – or better yet, specific recommendations. As Udstuen put it, “For us, AI means that there’s a recommendation produced. It’s not just an insight, it’s an action. And that action can be taken autonomously, or that action can be taken as a bit of advice that an operator can either take or not take.”
So what does this look like in practice? Udstuen painted a picture of an operator supervising a food production process. Rather than manually adjusting dials and settings, the operator interfaces with an AI model that automatically optimizes the process parameters. Critically, those AI recommendations are seamlessly integrated into the same screens and workflows the operator is already using.
Scaling this across an entire factory or enterprise requires a few key ingredients beyond the digital twin. One is easy connectivity to the myriad data sources on the plant floor, from historians to PLCs to SCADA systems. Another is empowering subject matter experts to act as “citizen developers,” using low-code and no-code tools to rapidly build and deploy new applications.
If you’re doing a digital transformation project, you can’t wait for a new thing to be created just to connect the data, that has to be an automatic thing. And then on top of that, getting the subject matter experts, the people that work in the plant that have the domain expertise, enabling those users to build applications in a no-code, low-code environment. That’s how you achieve scale.
Looking ahead, Udstuen sees generative AI – the hottest topic in tech right now – as just one part of a much broader industrial AI landscape. While he’s excited about the potential use cases, he stressed that manufacturers need to think holistically.
“If there’s one message I would like customers and prospects to hear, it’s that it takes considerably more to run a plant than generative AI, even as it’s fully developed,” he cautioned. “When we talk about industrial AI, it’s a superset of which generative AI is a subcomponent. You need to be thinking about running your plant with multiple flavors of AI, not just generative AI.”
For entrepreneurs looking to follow in Udstuen’s footsteps and build the next great industrial software company, he had one key piece of advice: a relentless focus on the customer’s pain point and the economic value of solving it. In a world awash with shiny new technologies, that clarity of purpose has never been more important.
To hear more of Erik Udstuen’s insights, check out the full episode of Augmented Ops wherever you get your podcasts to learn how TwinThread is helping manufacturers harness the power of industrial AI, visit twinthread.com.
