This paper develops a baseline agent-based macroeconomic model and contrasts it with the common dynamic stochastic general equilibrium approach. Although simple, the model can reproduce a lot of the stylized facts of business cycles. The author argues that agent-based modeling is an adequate response to the recently expressed criticism of macroeconomic methodology because it allows for aggregate behavior that is more than simply a replication of microeconomic optimization decisions in equilibrium. At the same time it allows for absolutely consistent microfoundations, including the structure and properties of markets. Most importantly, it does not depend on equilibrium assumptions or fictitious auctioneers and does therefore not rule out coordination failures, instability and crisis by definition. A situation that is very close to a general equilibrium can instead be shown to result endogenously from non-rational micro interaction.