The paper proposes an agent system for distributed generation setup identifying role and requirements of each genetic algorithms, due to their applicability for optimization solutions of nonlinear and stochastic scenarios, can find place in development of agents for distributed control. Advantages and special features of genetic algorithms far optimization are discussed, identifying some input and output variables. Requirement of compensator as an agent/sub-agent is also proposed. The state machine model/event loop model is proposed showing behaviour of system in real time scenarios. To facilitate understanding, object model of the agent system has also been suggested, correlation with state/event loop model where possible