The crystalline sponge (CS) method utilizes a crystalline porous material to arrange target molecules within its periodic pores. This enables the determination of the 3D atomic structures of organic molecules without the need for crystallization. However, its applicability is currently limited by the availability of suitable porous single crystals that can grow to a sufficient size for X-ray diffraction analysis. Although three-dimensional electron diffraction (3D ED) allows structure determination from nanosized crystals, ab initio structural analysis of organic molecules hosted in nanocrystalline sponges remains challenging and largely manual. Here, we present a 3D ED-based nanocrystalline sponge (NanoCS) workflow that integrates guest soaking, low-dose cryogenic data collection, and automated structure solution and refinement. A key advance is a newly developed automated approach for guest identification and structural analysis implemented in the AutoSolveX pipeline. Using the nanocrystalline bismuth-based metal–organic framework (MOF) SU-100 as a prototype crystalline sponge, we demonstrated the general applicability of this NanoCS strategy. 10 organic molecules, introduced as pure liquids, solutions, or vapors, are investigated. For all systems, 3D ED data collected under low electron fluence and cryogenic conditions enabled fully automated identification and refinement of the guest molecules using AutoSolveX. The results confirm the periodic arrangement of the guest molecules within the pores of SU-100, mediated by coordination bonding, hydrogen bonding, offset π–π stacking, and van der Waals interactions. This work establishes NanoCS combined with automated structural analysis as a practical and high-throughput platform for routine ab initio structural determination of organic molecules from nanocrystalline hosts.