Two recent developments have sparked discussions about AI agents and their broader implications. OpenClaw, an open-source project, allows users to grant AI extensive access to their digital lives, enabling it to perform tasks like sending emails and even negotiating car purchases. Shortly after, Moltbook emerged, creating a social network for these agents, where they interact and learn from each other in an unregulated environment. While this has generated excitement and media buzz, it also highlights the chaotic and often nonsensical nature of AI interactions. The rise of Moltbook raises important questions about how to design AI systems that operate as a collective rather than just individual tools. The article emphasizes the need for designers to consider the systemic effects of deploying swarms of agents. These agents can behave like groups of workers, collaborating on tasks but also competing for resources. This competition can lead to unpredictable outcomes, as seen with services like Google Maps or inventory bots that disrupt traditional markets. Designing for these interactions requires a new approach. It’s not enough to focus solely on individual user goals; the overall health of the system must also be a priority. Agents need to be aware of system constraints and adopt ethical behaviors that align with broader objectives. The challenge lies in instilling this awareness in agents while navigating the complexities of competing interests. The goal should be to foster a healthier ecosystem where agents act responsibly and contribute positively to the systems they inhabit.