Project NANDA aims to give AI agents an open infrastructure layer comparable to what the web gave documents. Ramesh Raskar describes a future with enormous numbers of agents that need portable identity, discovery, trust and commerce across organizationsInteroperability is the ability of different systems to exchange information and use it with compatible meaning and behavior. rather than proprietary directories and orchestration systems.
Maria Gorskikh explains how the NANDA Index resolves an agent identityAn AI agent identity is a persistent, verifiable identifier used to attribute messages, actions, permissions, and history to an agent. into a signed AgentFacts record, reachable endpoints and a guarded message box. Resolution can adapt to the requester, location and access rights, while businesses or individuals can publish through their own domains or a hosted service.
NANDA Town provides an open simulation environment for testing discovery, messaging, negotiation, payments and coordination at scale. Scripted or model-driven agentsA multi-agent system contains multiple AI agents that interact, coordinate, divide work, or influence one another while pursuing tasks. can run marketplace, auction, voting, consensus and supply-chain scenarios so protocol failures can be observedEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria. before the agent web becomes load-bearing infrastructure.
Watch on YouTube
