Researchers release a system for coordinating AI agents at web scale — first infrastructure for agents that interact autonomously

A team has released Holos, software infrastructure designed to let many AI agents coordinate with each other at scale without constant human oversight. This means the technical bottleneck that has kept AI agents isolated and dependent on human direction is now — in theory — removed, which is the basic condition required for any autonomous multi-agent system to actually work.

Until now, AI agents have mostly been solo task-solvers that work in isolation. This is a structural constraint, not a limitation of the models themselves. Holos attempts to remove that constraint by providing the plumbing that lets agents coordinate autonomously — pricing mechanisms, work assignment, value exchange — the way independent contractors or markets do. Whether this actually works at scale is unknown. But if it does, it's the infrastructure difference between AI as a tool you ask for things and AI as a system that operates independently and recruits other AI systems to do work.