The development of modern agent-based AI systems such as Claude Science, Co-Scientist and Robin presents new opportunities for research. The workshop will focus on the possibilities for designing AI agents across scientific disciplines, ranging from methodology (agent orchestration, benchmarking) to fundamental concepts (e.g. knowledge graphs), their implementation on HPC clusters, the security of agent-based systems (guardrailing, sandboxing) and the current opportunities available to researchers at Charles University. The workshop will also include discussions, mutual learning on the use of AI agents, and the sharing of experiences with these systems. If you are interested in contributing to the workshop with a presentation, please complete the form at this link; we will contact you once we have assessed the suitability of your proposal. The number of speakers is limited.
