DNS-AID vs ANS: How the Web May Identify and Discover AI Agents

Both aim to let systems determine what an agent is, what it can do, and whether it can be trusted.

DNS‑AID: Publishing Agent Capabilities in DNS

DNS‑AID, advanced by Infoblox, focuses on making AI agents discoverable.

The proposal defines a structured DNS namespace and record model that organizations can use to publish metadata about their agents—such as endpoints, supported protocols, or capabilities—within their existing domain infrastructure .

That means an organization could advertise information about an agent directly through DNS records associated with its domain. Other systems could then resolve those records to learn how to interact with that agent.

Key goals of DNS‑AID include:

• scalable discovery of AI agents across networks
• standardized metadata exchange between agents
• interoperability without requiring a central directory

Because DNS is globally distributed and already used for internet service discovery, proponents argue it provides a natural foundation for agent discovery at internet scale .

ANS: A Trust and Identity Layer for AI Agents

While DNS‑AID focuses on discovery, Agent Name Service (ANS) focuses on identity and verification.

ANS defines a system that anchors an AI agent’s identity to a domain name controlled by an organization, creating a verifiable trust relationship. The protocol uses public‑key infrastructure and certificates to bind an agent identity to that domain .

In practice, this means:

• an organization proves it controls a domain
• the agent identity is cryptographically linked to that domain
• other systems can verify the identity chain before interacting with the agent

The ANS draft describes the goal as providing a domain‑anchored trust layer for autonomous AI agents, particularly when agents operate across organizational boundaries where no single platform provides a universal trust system .

Why These Standards Are Built on DNS

Both proposals intentionally rely on DNS rather than creating a new naming or registry system.

DNS already provides:

• a global hierarchical naming system
• distributed governance
• wide deployment across the internet
• existing security features such as DNSSEC

Because organizations already control their own domains, they can publish agent identity and discovery information directly within their DNS infrastructure. That lets verification trace back to domain ownership, a model the internet already uses for email authentication, TLS certificates, and service discovery .

In other words, if a company controls example.com, it can publish and verify AI agents under that domain without depending on a centralized registry run by another platform.

Why the Work Is Being Proposed Through the IETF

DNS‑AID and ANS are currently being advanced as Internet‑Drafts within the Internet Engineering Task Force (IETF) standards process.

Internet‑Drafts are early proposals that allow engineers, infrastructure providers, and researchers to review and improve a protocol before it potentially becomes a formal internet standard.

Using the IETF process matters because:

• the specifications are publicly visible and open to industry review
• multiple vendors can implement compatible systems
• the resulting infrastructure can become part of the open internet rather than a proprietary platform feature

However, these documents remain draft proposals and are not yet official standards, so their final design and adoption remain uncertain .

Preventing a Single Vendor From Controlling the Agent Ecosystem

One of the motivations behind these efforts is to avoid a future where a single AI platform controls how agents are named, discovered, and verified online.

If discovery and identity were handled through proprietary agent marketplaces or centralized registries, platform operators could effectively decide:

• which agents appear in search or discovery
• which identities are trusted
• how agents interact with each other

By contrast, DNS‑based standards would allow any organization with a domain to publish and verify its own agents, much like hosting a website today.

That approach shifts authority from platform providers to domain ownership and open protocols, helping maintain interoperability across clouds, registrars, enterprises, and agent frameworks .

What Happens Next

DNS‑AID and ANS represent early infrastructure proposals for what some developers call the “agentic web.”

If the standards gain adoption, they could form part of a broader stack for agent interoperability that includes communication protocols, trust systems, and access controls.

For now, both proposals remain experimental drafts. Their future impact will depend on whether major cloud providers, AI platforms, and infrastructure companies decide to implement them as the number of autonomous agents operating online continues to grow.