A new Apache 2.0 specification standardizes how AI agents find and verify each other. If it succeeds, the strategic value shifts from model quality to enterprise distribution.
On June 17, 2026, a coalition of 12+ platform giants—including Google, Microsoft, NVIDIA, Salesforce, ServiceNow, Snowflake, and Cisco—introduced the Agentic Resource Discovery (ARD) specification. By establishing an open, Apache 2.0-licensed standard under the Linux Foundation, these companies are effectively working to commoditize the discovery and verification of AI agent capabilities across organizational boundaries.
ARD functions as a neutral registry system. It utilizes a static manifest, ai-catalog.json, paired with a dynamic POST /search endpoint to allow systems to find and verify agents. Crucially, this specification is designed to be complementary to existing protocols rather than a replacement. It integrates with the Model Context Protocol (MCP) for tool invocation and A2A for agent-to-agent communication. By providing a standardized way to catalog these resources, ARD lowers the barrier to entry for interoperability, allowing enterprise systems to navigate a fragmented landscape of proprietary and open-source agents with greater ease.
The strategic divergence here is notable. While Anthropic and OpenAI are Platinum members of the Agentic AI Foundation (AAIF), they are conspicuously absent from the list of ARD launch partners. This absence highlights a growing tension in the industry. Anthropic’s MCP, which has seen significant adoption with approximately 97 million downloads by early 2026, has served as a primary tool-access layer. However, the emergence of ARD suggests that platform incumbents are wary of allowing any single entity to control the exclusive gateway to enterprise workflows. By backing a neutral standard, these platforms are positioning themselves to ensure that the discovery layer remains open, thereby preventing any single model provider from locking in the distribution channel.
This move reflects a broader realization among enterprise leaders: the long-term strategic value in the AI industry is migrating toward the distribution and workflow integration layer. As the industry analysis publication Zendoric noted, “Open standards like ARD, if they truly remain open and under neutral governance such as the Linux Foundation, reduce friction and commoditize infrastructure. When the plumbing becomes cheap and universal, value moves upward—toward what gets built on top.” For companies like ServiceNow, which expanded its AI Control Tower to all enterprise AI systems in June 2026, the goal is to capture the value generated by the orchestration of these agents within complex business processes.
If ARD succeeds in becoming the industry standard for discovery, the competitive moat provided by proprietary models may begin to shrink. When the infrastructure for finding and deploying agents becomes a commodity, the advantage shifts to the platforms that control the distribution of those agents within enterprise workflows. This is a direct challenge to the current model-centric paradigm. With roughly 51% of organizations already pursuing a hybrid approach that combines neutral and proprietary orchestration, the market is clearly signaling a preference for flexibility over vendor lock-in.
However, the path forward for ARD is not guaranteed. The specification is new, and its long-term impact remains uncertain. While it is designed to be compatible with existing protocols, there is a risk of fragmentation between AAIF-backed projects and ARD-backed initiatives, which could slow adoption across the enterprise sector. Furthermore, while ARD could dilute the exclusive role of MCP as a tool-access layer, it does not replace it. The success of this initiative will depend on whether the platform giants can maintain the neutrality of the Linux Foundation governance model while navigating the competing interests of their own proprietary ecosystems.
Control over agent distribution is becoming the new strategic prize in the enterprise AI sector. As the industry matures, the ability to build the best model may become less critical than the ability to control the distribution of agents within the enterprise. By commoditizing the discovery layer, the ARD backers are attempting to rewrite the rules of engagement, ensuring that the future of AI is defined by the platforms that integrate these tools into the fabric of daily business operations, rather than the models that power them.