Kanzhi Cheng

Native desktop agents can finally bridge the gap between visual GUI interaction and programmatic code execution by learning a unified action space, achieving state-of-the-art performance on complex real-world tasks.

Effective visual tool use hinges on causally grounded supervision, not just imitating tool calls, reshaping how we train multimodal agents.

Even the best vision-language models struggle with reliable evaluation of computer-using agents, but OS-Shepherd models offer a low-cost solution that matches their performance.

OpenMobile proves that high-performing mobile agents can be trained on entirely synthetic, open-source data, closing the gap with closed-source models and enabling broader research.