Prime Intellect proposes mesh-style direct communication among AI agent swarms

Prime Intellect says agents should talk to each other directly, turning a tree-shaped structure into a mesh, so that no single context window has to hold all information.

In its account, context is distributed across the swarm and agents specialize over time, like a market where no one holds all the knowledge.

The company also says swarms carry costs: messages consume tokens, information can be lost, and agents must coordinate to avoid duplicated work. It is extending its RL training stack from individual agents to multi-agent systems, letting developers express arbitrary agent interactions and train them.