Llm swarms self-organize by learning one flexible local rule
Waggle: Learning One Anonymous Local Law for Self-Organizing LLM Swarms
Artificial Intelligence
Summary
As large language model (LLM) agents work together on complex tasks, deciding how they organize and communicate is important. The authors propose Waggle, a single local rule that all agents share and use, allowing them to coordinate without needing fixed roles or a central plan. They also develop a method called Swarm-Consistent Distillation to learn this rule so that coordination can form and adapt even when the group size or conditions change. This approach keeps the agents working well together and can transfer to new situations without retraining.
What this means in practice
- •For ai system engineers: Design agent networks that adaptively coordinate tasks without predefined roles or global control by using a shared local policy for flexible interaction.
- •For robotics control teams: Implement decentralized swarm behaviors in physical robots that reorganize coordination dynamically based on local rules learned once and reused.
Authors
Mingxi Zou, Wei Zhu, Zhuo Wang, Langzhang Liang, Zhiwen Tang, Yinghui Xu, Zenglin Xu
Abstract
As LLM agents increasingly collaborate on complex tasks, how to organize their interactions becomes a central design question. Existing multi-agent systems typically learn or adapt explicit roles, hierarchies, routing policies, or communication topologies. We shift the learning target to a reusable local law that can be shared across interchangeable agents and adapt coordination as populations or interaction conditions change, without redefining a global organization. We introduce Waggle, a shared anonymous policy over bounded local views that jointly selects task actions, semantic communication, and local commitment updates. Repeated execution of the same law allows coordination to form, persist, and reorganize online without explicit roles or global topology. To learn this law across interchangeable agents and evolving coordination, we develop Swarm-Consistent Distillation (SCD), combining anonymous-orbit consistency with rollout-grounded prediction of the next local coordination field, with no added inference-time components. Across diverse coordination settings, the same learned law remains effective as populations and interaction budgets change, retains over 96% of substrate-specific oracle quality, and transfers without retraining; SCD further improves reorganization after counterevidence. Together, these results show that LLM-agent organization can emerge and adapt through repeated execution of a learned local law.