This paper proposes a novel liquid intelligent metasurface (LIM)-assisted downlink multi-user multiple-input single-output (MISO) system, wherein both the base station (BS) and the metasurface are respectively equipped with fluid antennas (FA) and liquid elements. Unlike conventional reconfigurable metasurface-assisted systems with static geometries, the proposed architecture enables joint electromagnetic and spatial reconfigurability by allowing both the FA-empowered BS (FAS) and LIM to dynamically adjust their small-scale positions in addition to beamforming and phase-shift controls. We formulate a sum-rate maximization problem that jointly optimizes the BS beamforming, LIM phase-shifts, and the positions of fluid antennas and liquid elements. The problem is highly non-convex due to coupling between variables, fractional expressions, unit-modulus constraints as well as spatial correlation functions. To address these challenges, we adopt alternating optimization and introduce auxiliary variables and employ Lagrangian dual and quadratic transformation, successive convex approximation (SCA) as well as the penalty convex-concave procedure (PCCP) to solve the respective subproblems. Simulation results have demonstrated that the proposed FAS-LIM architecture significantly outperforms benchmark methods employing conventional fixed metasurface and fixed antenna arrays in terms of various parameter settings.