Usable Agent Discovery for Decentralized AI Systems
This work addresses the challenge of agent discovery in decentralized AI systems, which is critical for scalability and reliability, but the results are incremental as they extend known overlay trade-offs to a two-level churn scenario.
The paper studies decentralized agent discovery under two-level churn (node and agent) in large-scale agentic systems. It finds that structured overlays (Kademlia) are more robust and efficient in stable and node-churn regimes, while gossip-based overlays (Cyclon+Vicinity) remain competitive and can be faster when service readiness dominates.
Large-scale agentic systems run on distributed infrastructures where many software agents share physical hosts and are discovered via peer-to-peer mechanisms. Discovery must handle node-level churn from failures and host departures and agent-level churn from demand-driven activation, deactivation, and state changes. Their interaction reshapes classic trade-offs between structured and unstructured overlays. We study decentralized agent discovery under this two-level churn, assuming nodes host multiple agents, overlays are structured or gossip-based, and agents switch between warm and cold states. Using Kademlia as a structured and Cyclon+Vicinity as a gossip baseline, we compare stable, node-churn-only, agent-cooling-only, and combined regimes to see when routing efficiency, resilience, and service readiness align or favor different designs. Structured overlays are more robust and efficient in stable and node-churn regimes, while gossip-based overlays remain competitive and can be faster when readiness dominates.