AI agents are transforming domains from code generation to autonomous research workflows, yet time series forecasting remains largely unexplored. In this talk, we unpack what AI agents are, how they function, and why they’ve become a central topic in multimodal AI.

We’ll revisit core ideas in agent architectures, trace their emergence alongside foundation models, and explore why forecasting presents a distinctive and promising frontier for agent-based design. This conceptual talk is grounded in the open-source forecasting ecosystem and aims to spark a broader conversation: Are we ready for forecasting agents? And if not, what would it take to be?