AI agents will have their own search engine, known as Agentic Resource Discovery (ARD), backed by major corporations including Google and Microsoft. The ARD standard will facilitate the discovery of tools, skills, and other agents for AI applications.
ARD aims to create an open specification for publishing, discovering, and verifying AI capabilities across the web. This initiative involves collaboration among several companies, including GoDaddy, Hugging Face, NVIDIA, Salesforce, ServiceNow, Databricks, Snowflake, GitHub, and Cisco. Google and Microsoft both announced the partnership through blog posts.
Ramanathan Guha, technical fellow at Microsoft, stated, “AI is only as capable as its wiring allows,” emphasizing that AI agents can only utilize resources that they have been specifically connected to. This limits their capabilities, creating a need for a dedicated search engine to locate available resources.
The ARD framework is designed to function as an app store for AI agents, addressing the current difficulties in resource discovery. AI agents often struggle to find necessary tools and skills due to their limited connectivity to resources.
Microsoft compared the current situation to the early days of the web, where search engines like Yahoo relied on human indexers to create directories of websites. Google stated in its blog post, “Just as the open web democratized information, ARD democratizes AI resource discovery.” This suggests a shift toward more efficient resource utilization for AI applications.
However, the ARD announcement notably does not include OpenAI or Anthropic among its founding partners, marking a significant exclusion in the development of this new standard.
The introduction of ARD follows the earlier Model Context Protocol (MCP) from Anthropic, which established a method for AI systems to communicate with servers. MCP was seen as a foundational step, but ARD is positioned to provide a more comprehensive solution for AI resource discovery.
As AI agents become more reliant on diverse resources spread across various platforms, ARD is expected to enhance their operational capabilities by enabling more effective searches for needed tools and connections.
