Godela

Godela builds AI models of physical systems. A team uploads its own simulation data, for example a set of computational fluid dynamics runs, and Godela trains a physics constrained model on the behavior of that system. Engineers can then ask what if questions about airflow, geometry or materials and receive answers the vendor describes as solver level accuracy in seconds rather than days, and run optimization to converge on a best configuration. The site lists applications in data center thermals, electronics cooling, aerodynamics, electromagnetics and structures, where the model predicts stress and failure without running finite element analysis. It is aimed at engineering and research teams working on physical design problems who want to explore many configurations quickly. Workflow in the product has three steps: train a model from simulation data, explore it with plain language what if queries, and optimize toward a target such as the lowest peak temperature while holding total airflow constant. Ranked configurations are returned so engineers can compare designs by a metric such as cooling efficiency. Access is through a pilot request on the site, and no pricing is published. The company states it is part of Y Combinator's X25 batch.