Dr. Steven Lainé - SimScale - Transformig Simulation Workflows with AI Agents and Physics AI скачать в хорошем качестве

Dr. Steven Lainé - SimScale - Transformig Simulation Workflows with AI Agents and Physics AI

Short summary In this Future of Engineering Summit 2025 session, Dr. Steven Lainé (‪@SimscaleSimulation‬) discusses how AI agents and physics AI can eliminate simulation bottlenecks and significantly accelerate the design iteration process. He highlights the discrepancy between expectations and reality in the adoption of engineering AI, and demonstrates why cloud-native platforms are better suited to clean data and advanced AI programmes. A live demonstration shows how SimScale's in-platform simulation agent can set up and run analyses straight away with domain-specific customisation. Full description In this session from the 2025 Future of Engineering Summit, Steven Laines from SimScale discusses how simulation workflows are being transformed by physics AI and AI agents, building on the themes of generative design and agent-based automation that were covered throughout the event. Steven begins with a striking industry observation: although most engineering leaders anticipate productivity improvements from AI, only a few are currently achieving significant results. The biggest blockers? Siloed data and legacy tools, particularly in on-premises CAE environments. SimScale argues that cloud-native organisations are far more likely to have mature AI programmes and the centralised, clean data needed to make AI work in practice. He then links this to the growing 'need for speed': shorter RFP timelines, faster product release cycles and constant pressure to innovate without reducing margins. In traditional workflows, handoffs between simulations create bottlenecks in lead time — meaning teams may only manage one or two design iterations before deadlines. SimScale's solution is a combination of: Physics AI (inference) to deliver rapid predictions, enabling designers to swiftly explore thousands of design possibilities. Engineering AI (agents and automation) orchestrates end-to-end workflows, from understanding RFP requirements with LLMs to running simulations and feeding the results back into models and decision-making processes. Steven demonstrates this live on the SimScale platform: an out-of-the-box simulation specialist agent sets up a frequency analysis for a robotic gripper, assigns materials and constraints, runs the simulation and publishes the results in business tools. Steven then shows how agents can be customised with standard operating procedures (SOPs) and domain context (e.g. a valve agent for CV calculations) to reduce back-and-forth communication and automate standard workflows, which can be triggered from other systems such as Synera, or even tools like Microsoft Teams.