How Dassault Systèmes Is Building AI That Understands Physics - Ep. 296

Nicolas Saricierre, Vice President of 3D Experience Platform R&D at Dassault Systèmes, discusses the integration of AI into the 3D Experience Platform, which supports brands like CATIA and SolidWorks and serves 15 million scientists and engineers. A significant innovation is the development of virtual twins, detailed representations of products that allow for testing in a virtual environment before physical production.

Nicolas emphasizes the transformation of the platform from a SaaS model to an agent-as-a-service framework, embedding AI grounded in scientific principles and industry knowledge, distinguishing it from traditional generative AI. The technical pillars of Dassault's industry world models include industrial knowledge, which incorporates standards and engineering rules, and world understanding, focusing on specialized AI models that operate on virtual twins. These models ensure scientific accuracy by combining data systems, modeling, and simulation technologies.

Virtual companions, such as Leo the business expert and Marie the scientist, augment human workers, enhancing productivity while maintaining human oversight in decision-making. Trust in these companions is built on their scientific background and the management of intellectual property through IP Lifecycle Management. The integration of NVIDIA technologies enhances the capabilities of these AI models.

Dassault employs a hybrid approach to model development, utilizing both proprietary and best-in-class models, ensuring compliance with global regulations and focusing on interoperability through open standards. An example of practical application is Leo Mechanical Designer, which optimizes designs based on 3D scans or 2D drawings.

The transition to an agent-as-a-service model is driven by AI advancements, encouraging customers to address core business challenges to effectively leverage AI. The introduction of Leo has automated the generation of 3D parts, showcasing significant advancements in AI technology, including the evolution of virtual twins and companion simulations to enhance operational efficiency.

The SO system strategy aligns with NVIDIA's initiatives regarding AIQ and autonomous agents, addressing industrial challenges such as security and compliance. Virtual companions monitor factories, project execution, and supply chains in real-time, optimizing operations autonomously. Agents utilize virtual twins as training environments, conducting millions of simulations to provide validated solutions for engineers, leading to the evolution of virtual twins into self-improving assets over time. Increased communication about AI advancements is emphasized, particularly through collaboration with NVIDIA, to keep stakeholders informed about developments in the 3D experience platform.