This would be an amazing feature. We currently can't utilize any multi-modal functionalities within an AI Agent. It would be great to be able to pass ...