Cells are the fundamental units of life, functioning as intricate molecular machines, with organelles serving specialized roles within them. Among these, endosomes and lysosomes—collectively known as endolysosomes—play essential roles in intracellular transport, degradation, and signaling. Their movement within the cytoplasm follows a complex “stop-and-go” pattern, and disruptions in this dynamic behavior are linked to numerous diseases. However, the mechanisms governing this motion have long remained poorly understood. Recently, researchers from the Computational Biology and Machine Intelligence (CBMI) team at the National Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, leveraged deep learning-based image analysis to systematically uncover the dynamic patterns of endolysosome movement and their local microenvironmental context. Their findings reveal that nodes in the endoplasmic reticulum (ER) network act as critical regulatory hubs that control the switching of endolysosome motion states. This breakthrough demonstrates the transformative power of artificial intelligence in deciphering complex cellular dynamics and offers a new perspective on how organelles coordinate their spatial functions. Traditional cell dynamics research has relied heavily on manual annotation and trajectory analysis, which are time-consuming and ill-suited for handling high-throughput data or capturing the complexity of organelle movement under pathological conditions. In contrast, the AI-driven approach developed by the CBMI team enables automatic reconstruction of thousands of endolysosome trajectories, extracting detailed motion states, spatial distributions, and microenvironmental features. This allows for a comprehensive, high-resolution view of cellular dynamics. By integrating single-particle tracking, spatial distribution analysis, and ER morphology assessment, the team established a fully automated image analysis pipeline capable of multi-modal integration and precise interpretation of massive organelle dynamics data. This marks a pivotal shift from mere observation of motion to a mechanistic understanding of cellular regulation. The findings show that endolysosomes exhibit three distinct motion states within the ER network: global rapid movement, localized slow movement, and transient pauses. These pauses are frequently associated with transient interactions between endosomes and lysosomes, as well as organelle fission events, indicating a finely tuned coupling between the cellular cytoskeleton and the endomembrane transport network. This discovery highlights the ER network’s role as a dynamic “traffic hub” for intracellular organelle coordination. The AI tools developed in this study will be incorporated into the Panstone Scientific Foundation Model and the Panstone Digital Cell Platform, both deployed by the Chinese Academy of Sciences. These platforms aim to accelerate the translation of biological insights into clinical applications. The work exemplifies how artificial intelligence is becoming a core driver in life science discovery, enabling researchers to extract mechanistic knowledge from vast imaging datasets. The study was led by Dr. Wenjing Li, a deputy research fellow at the Institute of Automation, CAS, with Professor Ge Yang as the corresponding author. Key collaborators include Professor Junjie Hu from the Institute of Biophysics, CAS. The research was supported by the National Natural Science Foundation of China (Major Research Plan), the CAS Strategic Priority Research Program, the National Key Research and Development Program, and the Central University Basic Research Fund. Publication details: Endoplasmic reticulum junctions serve as a platform for endosome-lysosome interactions through their stop-and-go motion switching, Science Advances, Volume 11, Issue 38, September 2025. Full article link: [Insert link] Panstone Scientific Foundation Model access: [Insert link]