Google, Microsoft Launch ARD Standard for AI Agents as 11 Partners Back Open Discovery
Updated
Updated · ZDNet · Jun 18
Google, Microsoft Launch ARD Standard for AI Agents as 11 Partners Back Open Discovery
3 articles · Updated · ZDNet · Jun 18
Summary
Google and Microsoft said ARD will let AI agents discover, verify, and use tools, skills, and other agents across the web instead of relying only on pre-wired connections.
ARD tackles that discovery gap with 2 core pieces: catalogs—published as ai-catalog.json files on company domains—and registries that crawl and index those catalogs like search engines for the Agentic web.
11 partners including NVIDIA, Salesforce, ServiceNow, Snowflake, GitHub, Cisco, GoDaddy, Hugging Face, and Databricks joined the effort, while OpenAI and Anthropic were not listed.
GitHub, Hugging Face, and Google already showed reference implementations, and the Apache 2.0 specification is available now through AgenticResourceDiscovery.org and GitHub.
The same domain-anchored design that enables trust also creates a new attack surface, making compromised domains, DNS, repositories, or deployment pipelines high-value targets despite added trust metadata and enterprise controls.
Will standardizing AI discovery create new monopolies or truly democratize access to powerful AI capabilities?
When autonomous agents cause disasters, who becomes legally responsible for their choices in this new ecosystem?
The Agentic Resource Discovery (ARD) Standard: A New Foundation for Secure, Scalable, and Open AI Ecosystems
Overview
The Agentic Resource Discovery (ARD) standard is a major advancement in artificial intelligence, designed to democratize access to AI resources. By addressing the fragmented and complex process of finding and integrating AI models, tools, and data, ARD makes these resources more accessible and usable for everyone. It establishes a common framework that streamlines discovery, helping developers and organizations overcome hurdles in leveraging AI’s full potential. This standardization not only simplifies resource integration but also fosters innovation, paving the way for a more open and interconnected AI ecosystem.