Descriptions:
Thomas von Tschammer, co-founder and US managing director of Neural Concept, joins the Cognitive Revolution podcast hosted by Nathan Labenz to discuss how his Swiss company is using domain-specific AI models to transform physical product engineering. Neural Concept’s models can replicate results from expensive physics-based solvers in minutes rather than hours, enabling clients like Jaguar Land Rover to evaluate over 1,000 aerodynamic designs per day — a volume that was previously unthinkable.
The conversation covers Neural Concept’s engineering co-pilot product, which combines domain-specific prediction models with direct CAD platform interaction to make design changes autonomously. This agentic loop — optimization followed by domain-validated simulation — creates a natural structure for reinforcement learning. Tschammer also shares how Neural Concept found a niche in Formula 1, where regulations cap the compute teams can use for aerodynamic simulation, making per-run efficiency especially valuable. The discussion includes a frank exchange about which parts of the engineering process AI can already automate today versus where human judgment on cross-functional trade-offs remains essential.
The episode closes with a look at Neural Concept’s strategic trajectory: moving from training bespoke models per customer toward more general-purpose foundation models for engineering, and shifting pricing from seat-based to value-based models. For anyone tracking how AI is penetrating industrial and physical-world domains beyond software, this episode offers a grounded, practitioner-level view of what the pattern actually looks like in production.
📺 Source: Cognitive Revolution “How AI Changes Everything” · Published July 01, 2026
🏷️ Format: Interview
