block to establish numerous ingenious applications, such as supramolecular engineering, [1] nanotechnology, [2] soft robotics, and mechanical metamaterials. [3,4] When scaling down the size of each piece of the jigsaw puzzle and extending the concealed information to contain both light polarization and phase, the concept of optical jigsaw puzzles aiming for programmable optics can be obtained. It is known that the functions of traditional optical elements are endowed by a complex phase wavefront, e.g., an imaging lens has a parabolashaped phase distribution, and a grating has a periodic phase distribution, [5,6] and are expressed using the curved interface between different optical materials in geometrical optics. Metasurfaces, [7,8] composed of pixelated nanoantennas with different parameters of orientation, size, aspect ratio, spacing, etc., to obtain delicate control over the optical phase, polarization, or amplitude, have attracted tremendous scientific and industrial interest. However, these materials face the challenges of limited size, complicated microfabrication, and the dynamic manipulation of optical performance, which are the main obstacles to promoting the commercialization of smart, adaptive, and programmable photonics.Manual intervention in the self-organization of soft matter to obtain a desired superstructure is a complex but significant project. Specifically, optical components made fully or partially from reconfigurable and stimuli-responsive soft materials, referred to as soft photonics, are poised to form versatile platforms in various areas; however, a limited scale, narrow spectral adaptability, and poor stability are still formidable challenges. Herein, a facile way is developed to program the optical jigsaw puzzle of nematic liquid crystals via pixelated holographic surface reliefs, leading to an era of manufacturing for programmable soft materials with tailored functions. Multiscale jigsaw puzzles are established and endowed with unprecedented stability and durability, further sketching a prospective framework toward customized adaptive photonic architectures. This work demonstrates a reliable and efficient approach for directly assembling soft matter, unlocking the long-sought full potential of stimuli-responsive soft systems, and providing opportunities to inspire the next generation of soft photonics and relevant areas.