In a recent episode of the NVIDIA AI Podcast, Nicolas Cerisier, Vice President of 3DEXPERIENCE Platform R&D at Dassault Systèmes, revealed how the company is pioneering a shift from traditional Software-as-a-Service (SaaS) to "Agent-as-a-Service." This new paradigm introduces "Industry World Models" that harness the scientific laws of physics and engineering to power autonomous virtual companions for industrial applications.
"Industrial AI can do more than just observe," Cerisier explained. "By grounding AI in physics, we create models that truly understand how materials behave, how forces interact, and how systems fail—enabling predictive and prescriptive intelligence at scale."
The approach marks a significant departure from conventional AI that relies solely on pattern recognition. Dassault Systèmes' models integrate fundamental physics principles, allowing them to simulate real-world phenomena with high fidelity. This enables engineers to design, test, and optimize products virtually before physical prototypes are built.
Cerisier highlighted that these Industry World Models are already being deployed across sectors such as aerospace, automotive, and life sciences, where safety and precision are paramount. The shift to Agent-as-a-Service means companies can access these sophisticated AI agents on demand, reducing the need for in-house expertise and infrastructure.
As AI becomes more embedded in industrial processes, the need for physically accurate models grows. Dassault Systèmes' physics-grounded AI aims to bridge the gap between digital simulation and real-world performance, offering a new level of intelligence that understands not just data, but the laws that govern the physical world.