The 1st Workshop on
Agentic Multimodal
Intelligence

Recent advances in multimodal foundation models—including large vision-language models and multimodal generative models—have significantly reshaped artificial intelligence by enabling unified perception, understanding, and generation across text, images, videos, audio, and other sensory modalities. Beyond perception and generation, a growing body of research indicates that the next frontier of intelligence lies in agentic AI: systems that can autonomously reason, plan, act, and adapt over time in interactive environments. This shift marks a fundamental transition from passive multimodal models to Agentic Multimodal Intelligence (AMI).

This workshop brings together researchers and practitioners across multimedia, vision, natural language processing, robotics, and human-computer interaction to advance agentic intelligence—fostering interdisciplinary discussion on models, data, benchmarks, and evaluation for next-generation intelligent agents. We particularly encourage work that bridges multimodal learning, reinforcement learning, embodied AI, and human-in-the-loop systems toward building scalable, robust, and trustworthy agentic systems.