How Dassault Systèmes Is Building AI That Understands Physics | NVIDIA AI Podcast Ep. 296
Summary
Dassault Systèmes, a company with 400,000 customers and 45 million users, is transforming its 3DEXPERIENCE platform from a SaaS architecture to an agent-as-a-service model, integrating AI into its core. This shift enables the creation of "Virtual Twins"—scientific, multidisciplinary, multiscale representations of products—that can be tested virtually before physical existence. The company is developing "Industry World Models" grounded in science, fueled by industry knowledge, and sovereign by design, which understand scientific foundations like physics and engineering rules. These models power "Virtual Companions" like AURA (business expert), LEO (engineer), and MARIE (scientist), which act as AI coworkers to augment human capabilities, orchestrate actions, and comply with regulations. Dassault Systèmes leverages NVIDIA technologies, including NIMs models, Nemotron 3 Super, and AI-Q Blueprint, across understanding, reasoning, and execution layers, maintaining a 25-year partnership to accelerate industrial AI.
Key takeaway
For AI Engineers and Research Scientists developing industrial applications, consider how agent-as-a-service platforms can integrate scientific foundations and industry-specific knowledge to create reliable, trustable AI. Your focus should be on building systems that augment human capabilities and ensure auditability, traceability, and compliance, rather than merely adding AI on top of existing solutions. Explore how Virtual Twins can become self-evolving assets, continuously optimizing processes without constant human prompting.
Key insights
Dassault Systèmes is transitioning to an agent-as-a-service platform, using AI-driven Virtual Twins and Companions for industrial innovation.
Principles
- AI for industry must be grounded in science.
- Industrial AI should be fueled by industry knowledge.
- Systems must be sovereign by design.
Method
Industry World Models combine scientific, physical laws with multi-scale modeling and simulation, integrating industrial knowledge and specialized AI models. Virtual Companions then use this intelligence to execute actions within business contexts.
In practice
- Use LEO mechanical designer to generate optimized, manufacture-ready designs.
- Integrate NVIDIA NIMs models for seamless AI deployment.
- Leverage Nemotron 3 Super for improved reasoning performance.
Topics
- Dassault Systèmes
- 3DEXPERIENCE Platform
- Agentic AI Systems
- Industry World Models
- Virtual Companions
Best for: AI Architect, AI Scientist, AI Engineer, Research Scientist, Director of AI/ML
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by NVIDIA.