Building performance simulation intimately interacts with the architectural design process. The integration of building simulation with NURBS CAD modeling provides new approaches to satisfy high-performance goals in the design phase. Simulation tools are usually used to evaluate performance of building shapes and facades in the early stages of decision-making. Furthermore, these tools are integrated with optimization models that allow the discovery of better performing designs to the user. However, these performance tools are rarely used for space layout optimization due to difficulties in controlling indoor space geometry. Relationships between different space geometry require a significant number of variables to consider for optimization, and the most popular way to ease this problem is adapting a matrix format. Thus, geometrical conditions in space layout optimization have been restricted to orthogonal 2D spatial shapes. This paper proposes a methodology for interior space optimization focused on an agent-based hierarchical control system to improve natural daylight for indoor settings. The optimization process contains three-dimensional analysis and visualization by the link between a NURBS CAD tool and simulation tools. The placing and sizing of interior spaces are controlled by a relatively small number of control points, also called agent points, and evaluates the performance of interior natural daylighting conditions. This study is to demonstrate the potential of a 3D-space layout with more flexible geometrical shapes and control with performance based on optimization.