Building the Open Agentic Web: Why Identity and Discovery Must Work Together

The internet was built around people. Websites, browsers, search engines and applications were all designed with human users in mind.

The next era of the internet will be different.

Increasingly, software agents will discover information, make decisions, complete transactions and interact with each other on behalf of people and businesses. As this transition accelerates, a new layer of internet infrastructure is emerging. One designed for agents discovering, evaluating and interacting with one another.

For that future to work at scale, agents need more than intelligence. Two fundamental questions emerge:

How do agents establish verifiable identity?
How do they discover the capabilities they need?

At GoDaddy, we believe identity and discovery are inextricably linked and equally important. We started with identity, collaborating with other industry stakeholders to develop Agent Name Service (ANS), an open framework for verifiable agent identity.

And today, we're proud to announce the Agentic Resource Discovery (ARD) Specification, an open specification for dynamic agent discovery, developed in collaboration with industry leaders including Cisco, Databricks, GitHub, Google, Hugging Face, Microsoft, Nvidia, Salesforce, SAP, ServiceNow and Snowflake. ARD is designed to standardize how AI agents discover and use tools, services and capabilities across the ecosystem.

ANS is designed to help agents, services and organizations establish domain-based identities so participants can validate who is providing a capability before engaging with it. In an agentic web, that kind of identity layer is foundational to accountability and interoperability.

Together, ANS and ARD address two foundational layers of the agentic web:

ANS is the identity layer: How agents, organizations and services establish verifiable identities.
ARD is the discovery layer: How agents search for and discover capabilities.

By making ANS catalogues ARD-compatible and exposing them through an ARD-compliant search interface, we help ensure that identity and discovery work seamlessly together across the ecosystem.

Neither layer is sufficient on its own. Discovery without identity creates uncertainty. Identity without discovery limits utility. Together, they provide the foundation for an open and interoperable agentic ecosystem.

Identity is foundational infrastructure

Identity becomes increasingly important as agents become more capable and autonomous.

When an agent discovers a service, requests information, initiates a transaction or acts on behalf of a user, participants need confidence about who is operating that agent and who stands behind the services it interacts with.

That’s the role of ANS.

Built on established internet principles, ANS enables agents and services to establish verifiable, domain-based identities. It allows participants to validate ownership and authenticate interactions between agent-to-agent and agent-to-resource communications.

In practice, this means that when an agent discovers a capability through ARD, it can verify the identity of the provider through ANS before engaging with it.

Just as domain names helped establish trust and accountability on the human web, agent identity will play a critical role in the agentic web.

The shift from static integrations to dynamic discovery

Today's agentic ecosystems largely rely on preconfigured integrations.

Developers decide in advance which tools an agent can access, manually connect those systems and maintain those integrations over time. As the number of agents, capabilities, services and tools proliferates, this approach becomes increasingly difficult to scale.

ARD introduces a different model.

Rather than relying on static catalogues or predefined integrations, agents can discover capabilities based on intent. When an agent needs to perform a task, it can search for relevant tools, services, MCP servers, A2A agents, ANS registered-resources or other AI-enabled resources and identify appropriate capabilities in real time.

This transforms discovery from a developer-managed process into a network capability.

For developers, it means building once and becoming discoverable across a growing ecosystem. For organizations, it means expanding what their agents can do without continuously rebuilding integrations. For the broader industry, it creates a path toward interoperability that can scale alongside innovation.

Interoperability by design

A core goal of both ARD and ANS is interoperability.

The future agent ecosystem will not be built around a single registry, marketplace or platform. Agents will need to discover capabilities globally across an ever-expanding number of providers, organizations, and networks.

That is why compatibility between identity systems and discovery systems matters.

Within ANS, agent metadata and capability descriptions can be published in a manner that aligns with the ARD specification. At the same time, ANS exposes a fully ARD-compliant search interface, enabling ARD-enabled agents and platforms to discover ANS-published resources using the same discovery patterns they use elsewhere in the ecosystem.

This approach allows organizations to benefit from identity through ANS while participating in a broader discovery ecosystem through ARD.

Put simply, ANS does not create a separate discovery universe. It helps extend the reach of the open agentic ecosystem by making identities and capabilities discoverable through shared standards.

Building on the momentum of open standards

The agent ecosystem is already benefiting from a growing set of open standards.

Protocols such as Model Context Protocol (MCP) have created common ways for agents to interact with tools and services. Emerging standards such as Agent-to-Agent (A2A) are helping establish patterns for agent collaboration and communication.

ARD and ANS complement these efforts by addressing a different challenge: identity and discovery.

If MCP helps define how agents use tools, ARD helps agents find those tools. If A2A helps agents communicate, ARD helps them discover potential collaborators. ANS, in turn, provides an identity layer that can operate across all of these interactions.

These standards are not competing approaches. They are complementary components of a larger ecosystem that is becoming increasingly interoperable.

An open standard, not a platform

One of the most important aspects of ARD and ANS is what they are not.

They are not products.

They are not marketplaces.

They are not controlled by any single company.

ARD and ANS are shared, open specifications that organizations can implement independently. Any agent, service, platform or tool provider can participate. The objective is to establish a common framework that enables it across a diverse ecosystem rather than centralize identity and discovery.

That distinction matters.

The internet succeeded because open protocols created a foundation on which companies could innovate. Organizations competed on products, services and customer experiences while benefiting from shared infrastructure underneath.

The same dynamic is emerging in the agent ecosystem.

The greatest opportunities will come from networks of interoperable agents, tools and services where open standards work together regardless of who built them.

Building the next layer of the internet

The transition to an agentic web will not happen because a single company builds the winning platform.

It will happen because the industry collectively establishes the foundational infrastructure that allows innovation to scale.

ANS represents an important step toward that future by providing the verifiable identity layer necessary for accountability and interoperability, while ARD complements that vision by enabling dynamic discovery across a growing ecosystem of agentic capabilities.

Together, they lay the groundwork for a more open agentic web. Developers build once and get discovered broadly. Agents access capabilities dynamically. Verification and accountability are built into the infrastructure itself.