AI-Assisted Simulation and Agentic Workflows

Name: AI-Assisted Simulation and Agentic Workflows
Uploaded: 2026-07-30 19:36:44
Duration: 50 min 56 s
Description: Watch this video where Bjorn Sjodin of COMSOL and Rui Aguiar of Cosmon talk about accelerating R&D with AI-assisted simulation and agentic AI workflows.

Back to Video Gallery

Duration: 50:56

Press O for more options

In this archived COMSOL webinar, Bjorn Sjodin of COMSOL and Rui Aguiar, CEO of Cosmon, discuss how AI tools and simulation software can be combined to speed up engineering work. Sjodin first reviews the AI functionality built into COMSOL Multiphysics®: the Model Builder together with its Java-based Application Programming Interface (API), and the Chatbot window, which connects to any OpenAI-compatible large language model and offers general, programming, and modeling context modes. He demonstrates using the chatbot alongside the Java shell window and the method editor to generate API code (for example, creating a cylinder), to find logical and mathematical errors in code, to answer general modeling questions, and to analyze an attached model or a screenshot of the desktop; he also notes current LLM limitations with spatial reasoning and very large mesh data sets, and mentions COMSOL's separate surrogate-model tools. He then shows how an external coding agent, such as OpenAI's Codex, can connect to COMSOL through its API and autonomously build a model, illustrated with a geodesic surface-distance example. Aguiar then explains agentic workflows — the plan, act, and observe agent loop and how it differs from traditional machine-learning models and chatbot co-pilots — and walks through simulation use cases where agents save time: pre-processing (turning a 20-page specification into a meshed bracket thermal model), troubleshooting solver non-convergence, post-processing with automated report generation, and design-of-experiments loops. He cites roughly a 2–3× faster time-to-completed-model and about 60% less manual effort, and frames a near-term shift in which engineers move from operators to directors of agent-driven workflows while keeping a human in the loop.

Overview
Transcript

In this archived webinar, Bjorn Sjodin of COMSOL and Rui Aguiar of Cosmon discuss how AI tools and simulation software can be used in tandem to accelerate productivity and optimize workflows. The webinar begins with an introduction by Sjodin that outlines the AI functionality in the COMSOL Multiphysics^® software — including the COMSOL API and the Chatbot window, which can assist with modeling tasks, generate and correct COMSOL API code, and help automate workflows. From here, Aguiar does a deep dive into how AI agents can reduce repetitive work, streamline simulation and design workflows, and help engineering teams move more efficiently through every step of the product development process.

Rui Aguiar is an engineer, researcher, entrepreneur, and the CEO of Cosmon, where he focuses on the next frontier of software development by building advanced, agentic tools for engineers.

Chapter Selection

AI Functionality in COMSOL Multiphysics. (2:13) AI Agents (12:40) Agentic Use of COMSOL Multiphysics^® (13:59) Introducing NEXUS (22:17) NEXUS in the Simulation Workflow (25:56) DEMO: From Paper to Simulation (27:31) Nexus for Troubleshooting & Convergence (31:54) DEMO: Autonomous Troubleshooting (33:12) Nexus for Postprocessing & Reporting (36:38) DEMO: Postprocessing and Report Generation with Nexus (38:25) Nexus for Design Loops & Optimization (43:09) The Future of Agentic AI in Engineering (45:23)

To download the files, please sign into your COMSOL Access account.*