How it works
NandaIndex is an AI Catalog-formatted database of index records. It does not introduce a new record format. Each entry points to the correct next discovery object for a given resource identifier.
my.nandaindex.org
Federated resolution
Federated resolution separates resolution from discovery. The proposed architecture reuses AI Catalog as a unified record format for resolution. Each resolution entry is an AI Catalog-compatible record that acts as a typed pointer.
- A requester starts with a lookup key or subject identity, for example, a domain, email address, DID, or AIR URN.
- The requester queries a resolution system.
- The system returns AI Catalog-formatted resolution entries.
- Each entry specifies a discovery path.
- The requester follows that path through ARD, DNS lookup, a gateway, or another native mechanism.
- Standard discovery, verification, and invocation proceed.
Resource identifier
A resource identifier is a stable identifier used to initiate resolution, such as a domain, DID, or provider-verified account identifier.
Resolution
Resolution determines where and how discovery should begin. The resolution layer maps resource identifiers to the appropriate discovery entry point.
Discovery
The discovery layer defines how agents search for relevant capabilities at runtime. The Agentic Resource Discovery (ARD) specification provides a standard interface for querying and retrieving agentic resources across catalogs and registries.
Invocation
Invocation is interaction with the selected resource through A2A, MCP, REST, or another supported protocol.
Deployment contexts and use cases
The federated resolution architecture applies across heterogeneous environments with different infrastructure, identity, and governance models.
Enterprise AI Catalog
For enterprises that publish AI Catalog at a well-known endpoint and own the domain, direct resolution already works. NandaIndex is not in the critical path. A NandaIndex entry remains useful for fallback, federation visibility, migration, and anti-squatting.
Enterprise DNS-based service discovery
SkyBlue uses DNS-based service discovery via SVCB. NandaIndex makes this DNS-based discovery path reachable from the global switchboard while representing the target using a recognized A2A Agent Card type.
SMB: Moon Bakery
Moon Bakery owns moonbakery.com but runs no enterprise infrastructure. Its runtime, agent card, and domain are with three separate providers. Moon Bakery needs only a stable resource identifier and a delegated path. NandaIndex can provide a resolution entry point for SMBs.
Individual: [email protected]
John has no domain. His runtime is on Azure; his agent card is with a third-party host. No personally controlled domain is required. NandaIndex enables identity-first discovery for individuals, when the underlying account identity is verifiably bound to the resolution record.