In this paper, we present an agent-based system to control a single robot's behaviour. We present an artificial genome structure, based on gene regulatory networks, in which several regions can be distinguished such as promoter regions, indicator genes, transcription factor binding sites, regulatory genes and expressed genes. We use agent-based modeling (ABM) to simulate a bio-inspired system based on the artificial genome, with the ultimate goal of providing phenotypic information for a simulated robot. We show that the presence of a feedback loop in the agent-based system, along with the corresponding agent replacements, is essential to allow the robot to perform its tasks.