Skip to content

NANDA Index

NANDA research explores the infrastructure required for an open Internet of AI Agents.

The work examines how autonomous agents can discover, communicate, coordinate, establish trust, and operate across different systems and organizational boundaries.


Research Areas

Agent Discovery

How agents and agentic resources can be discovered across different registries, platforms, and ecosystems.

Resolution

How stable agent identities can be connected to appropriate discovery and communication mechanisms as agents and their environments change.

Interoperability

How independently developed agent systems can communicate and work together without requiring a single framework or protocol.

Trust

How identity, capabilities, provenance, credentials, and other claims about agents can be verified.

Coordination

How populations of agents can coordinate decisions and actions in decentralized environments.

Evaluation

How individual agents and agent populations can be evaluated under realistic operating conditions.

Governance

How technical infrastructure, institutional responsibilities, and shared standards can support agents operating across organizational and jurisdictional boundaries.


Research Work

NANDA research is developed through specifications, technical papers, prototypes, experiments, and open proposals.

Research currently includes work on:

  • Agent interoperability
  • Agent coordination
  • Agent evaluation and sandboxes
  • Trust and verification
  • Governance of the Agentic Internet
  • Agent infrastructure and resolution

Publications

Research papers and technical publications produced through NANDA and its collaborators:


Specifications

NANDA research also produces technical specifications and implementation proposals:


Open Questions

The Internet of AI Agents is still developing. NANDA research continues to examine questions around:

  • How should agents be identified and discovered at scale?
  • How should capabilities and claims be represented and verified?
  • How should independent agent ecosystems interoperate?
  • How should agents coordinate without centralized control?
  • How should agent behavior and network-level performance be evaluated?
  • How should autonomous agents operate across organizational and jurisdictional boundaries?