Veeda AI builds multimodal foundation world models: scalable simulations of physical reality where embodied agents learn through interaction.