The latest advances in the gridbrain agent brain model are presented. The gridbrain models brains as networks of computational components. The components are used as building blocks for computation, and provide base functionalities like: input/output, boolean logic, arithmetic, clocks and memory. The multi-grid architecture as a way to process variable sized information from different sensory channels is addressed. We show how an evolutionary multi-agent simulation may use the gridbrain model to emerge behaviors. The Simulation Embedded Genetic Algorithm (SEGA), aimed at continuous multi-agent simulations with no generations is described. An experimental scenario is presented where agents must use information from two different sensory channels and cooperate to destroy moving targets in a continuous physical simulation. Results are analysed and synchronization mechanism are shown to emerge.