In this paper, a shape sampling approach is proposed for CAD products that can be used to suggest innovative product shapes to designers and consumers. These shapes are intended to inspire designers and can be employed during the design process. For a given set of geometric parameters defining the product shapes, parameter relationships (i.e., geometric constraints), and parameter ranges, a particle tracing (PT) algorithm is proposed to find product shapes that satisfy the defined geometric constraints in the shape space. Particles are placed at points in the shape space by minimizing the Audze-Eglais potential energy of the particle positions using a permutation genetic algorithm. They then move until one of the predetermined stopping criteria is met. Particle movement is achieved using a cost function that favors movement towards feasible shapes. By iteratively running the PT algorithm, feasible shapes are obtained. Representatives of these shapes are identified using a kmedoids clustering approach, and such representatives can be used by designers or shown to consumers to customize the product according to their preferences. In this paper, eight CAD models (e.g., car hood, yacht hull, wheel rim) are utilized to validate the performance of the proposed sampling technique. We also compare our technique with related methods.