Skip to main content

Ant Group Open Sources Avernet: A New Infrastructure for Multi-Agent Collaboration

Ant Group's inclusionAI team has officially open-sourced Avernet V0.1, a new infrastructure designed to streamline multi-agent collaboration. As large language models become more prevalent, the challenge of getting multiple AI agents to work together efficiently on complex tasks has grown. Avernet aims to solve that by providing a standardized "collaboration layer" that handles registration, discovery, invitation, consensus, and traceable execution—without replacing the agents' own reasoning capabilities.

Image

The system supports various interaction modes, including free-form chat and leader-follower setups. By using groups, sessions, and shared context, it helps agents reach consensus more effectively. A built-in feedback loop allows the system to learn and improve over time through observation, evaluation, reuse, and optimization.

Avernet is designed to be compatible with existing ecosystems. It supports mainstream protocols like OpenClaw and works with custom agents, third-party engines, and bot platforms. Developers can get started quickly with Docker images or local deployment, lowering the technical barrier.

Industry observers note that while Avernet is still in its early V0.1 stage, it directly addresses common pain points in multi-agent cooperation, such as difficulty in discovery and disorganized collaboration. By turning collaboration logic into infrastructure, it offers a clear and efficient path forward for the industry.

Key Points:

  • Avernet is an open-source infrastructure for multi-agent collaboration, not a replacement for agent reasoning.
  • It standardizes processes like registration, discovery, invitation, consensus, and traceable execution.
  • Supports flexible interaction modes and includes a feedback loop for continuous improvement.
  • Compatible with mainstream protocols and existing agent ecosystems.
  • Available via Docker or local deployment for easy setup.