"By applying advanced analytics and AI at scale, we were able to see significant increases in fuel economy, productivity, as well as avoidance of major downtime events." - Erik Udstuen

‍

In this episode, Parthesh Shastri talks with Erik Udstuen, CEO of TwinThread, about the future of industrial AI and digital twins. Udstuen shares his journey from GE’s energy division to founding TwinThread, highlighting the blend of generative and traditional AI in industrial settings. He advocates for virtual operations centers that scale expertise globally—illustrated by a success story at Colgate’s pet food division—and outlines strategies for IP protection, including unidirectional communication, isolated language models, and secure, client-based cloud deployments. Udstuen sees industrial AI advancing from frontline support to product innovation, calling for safe experimentation as manufacturing faces a “triple squeeze” of constraints.

‍

‍

• Erik Udstuen’s experience at GE managing gas turbine fleets shaped his vision for industrial AI, improving fuel efficiency, productivity, and uptime across sectors.
• TwinThread’s AI system transformed quality control at Colgate’s pet food division, analyzing sensor data to shift from co-pilot support to full closed-loop automation.
• Udstuen defines industrial AI as a full-stack ecosystem—blending generative and traditional methods—requiring robust architecture for real-world integration.
• TwinThread ensures data security with one-way data flows, isolated language models, and client-hosted deployments, preserving control and functionality.
• Erik’s early work in neural networks for pulp manufacturing set production records; today, lower AI deployment costs unlock access across industrial sectors.
• Udstuen sees AI enabling fewer engineers to manage more sites via centers of excellence and predicts future applications in product design and material science—all within secure, IP-safe frameworks.