The paper examines the role of finite automata within the context of multi-agent systems. It is structured into two main parts: theoretical and practical. The theoretical section defines key concepts such as agents, their various architectures, and the rationale for integrating multiple agents into a cohesive multi-agent system. It emphasizes the importance of coordination, cooperation, and communication among agents to enhance system functionality. The paper also explores the relationship between agents and finite state machines, highlighting how finite automata can be utilized to model agent behaviors effectively. In the practical part, the authors present a specific design and implementation that combines the theoretical concepts discussed. They illustrate how subsumptive architecture can be employed to create behavior-based agents that operate in parallel, responding reactively to environmental stimuli. The paper also addresses the challenges of resource limitations in platforms like Arduino, suggesting optimizations for code modularity and communication protocols. Overall, this work contributes to the understanding of how finite automata can be leveraged to improve the design and functionality of multi-agent systems, paving the way for more sophisticated and efficient automated solutions in various applications.