AI Agents Just Got a Phonebook for the Entire Internet
On June 17, 2026, a coalition of 11 major technology companies — led by Google and Microsoft — published the Agentic Resource Discovery (ARD) specification, an open standard that gives AI agents a universal way to find tools, skills, and other agents at runtime. Think of it as the Domain Name System (DNS) for the agentic web — a discovery layer that finally lets AI software navigate the internet's sprawling ecosystem of capabilities without being explicitly wired to every single endpoint.
The announcement marks a pivotal moment for the AI agent ecosystem that has been rapidly reshaping how we interact with technology. Until now, agents could only use tools they had been pre-configured to access — a severe bottleneck that the industry has been wrestling with as agentic AI moves from demos to production.
Why ARD Matters: The Discovery Problem
AI agents today face three fundamental questions: Where does the right capability live? Which capability should I actually use? And how do I verify it's safe to connect to? The current landscape offers powerful protocols like MCP (Model Context Protocol, for how agents call tools) and A2A (Agent-to-Agent, for how agents talk to each other), but what's missing is a standard way to find capabilities across organizational boundaries and establish trust in what you find.
As Ramanathan Guha of Microsoft put it: "AI can only use what it's been explicitly wired to use. Everything else may as well not even exist." ARD solves this by sitting as a discovery layer in front of MCP and A2A, acting as the search engine for the agentic world.
How ARD Works: Four Phases
The specification, hosted under Apache 2.0 licensing and built on the AI Catalog data model maintained under the Linux Foundation, defines four core phases:
1. Publishing the Catalog
Organizations host a manifest file called ai-catalog.json at a well-known path on their own domain. This catalog lists available capabilities — MCP servers, A2A agents, OpenAPI tools, or even nested sub-catalogs. Because the file lives on the organization's own domain, domain ownership itself becomes the cryptographic foundation for identity and trust.
2. Discovery and Resolution
Registries — think of them as search engines for agents — crawl these catalogs and index their contents. When an agent needs a capability, it queries a registry using plain-language intent, or directly fetches a catalog from a known partner domain. This mirrors how payment networks like Mastercard have begun creating agent-to-agent payment infrastructure that similarly needs dynamic discovery.
3. Cryptographic Verification
Before connecting, the client agent cryptographically verifies the provider's true identity. Trust metadata is attached at the discovery layer, allowing confirmation before any runtime connection is established.
4. Direct Runtime Connection
Once verified, ARD steps aside. The agent connects directly to the capability using its native protocol — MCP, A2A, OpenAPI, or whatever else — with no further overhead from the discovery layer.
The Coalition and Immediate Implementations
The specification is co-developed by 11 companies : Cisco, Databricks, GitHub, GoDaddy, Google, Hugging Face, Microsoft, NVIDIA, Salesforce, ServiceNow, and Snowflake. Several partners shipped implementations the same day:
- GitHub released Agent Finder for Copilot , a reference implementation that lets developers describe a task in natural language and receive ranked matches from curated public catalogs or private enterprise registries.
- Hugging Face launched a Discover Tool providing semantic search across thousands of skills and MCP servers on its Hub.
- Google announced that its Agent Registry within the Gemini Enterprise Agent Platform will add native ARD support in the coming months, with enterprise-grade features like globally unique URNs, agentic egress policies, and compliance standards support including HIPAA.
The timing aligns with a broader industry shift: Qualcomm's CEO recently predicted that AI agents will replace traditional apps, and Salesforce's $3.6 billion acquisition of Fin underscored just how seriously the industry is taking agentic infrastructure.
What This Means for Developers and Businesses
ARD is currently a v0.9 draft , and the consortium is actively inviting feedback through the official GitHub repository. The project's website includes a quickstart guide for publishing your first catalog.
For developers, ARD means the end of hard-coded tool integrations. For businesses, it means their AI services become automatically discoverable without custom outreach. And for the broader ecosystem, it's the missing infrastructure layer that could finally make the vision of interoperable, cross-organizational AI agents a practical reality.
Source: Google Developers Blog | Technobezz
Originally published on TekMag
Top comments (0)