AR multi-agents