The article presents a model that combines fully rational and bounded rational agents to study monetary policy. They find that the Taylor condition is important for stability when the proportion of fully rational agents is fixed or changes through learning. However, if the policy includes interest rate smoothing, stability may be compromised. In such cases, the model shows the presence of limit cycles and a path to randomness. This research contributes to understanding how different types of agents can affect the effectiveness of monetary policy.