Circular geometry provides a rotating 'sweep signal' that stops AI agents from talking over each other.
April 23, 2026
Original Paper
Phase-Scheduled Multi-Agent Systems for Token-Efficient Coordination
arXiv · 2604.17400
The Takeaway
Multi-agent systems often waste tokens and time because coordination is messy and unorganized. This method uses concepts from algebraic topology to schedule agent actions on a circular manifold. Each agent only speaks when the rotating signal reaches its specific phase. It turns a chaotic group chat into a perfectly timed relay race. This reduces token waste and keeps the conversation focused on the goal. Geometry is becoming a powerful tool for managing the behavior of large AI swarms.
From the abstract
Multi-agent systems (MAS) powered by large language models suffer from severe token inefficiency arising from two compounding sources: (i) unstructured parallel execution, where all agents activate simultaneously irrespective of input readiness; and (ii) unrestricted context sharing, where every agent receives the full accumulated context regardless of relevance. Existing mitigation strategies - static pruning, hierarchical decomposition, and learned routing - treat coordination as a structural