Article illustration 1

As AI agents—autonomous programs that execute complex tasks without step-by-step instructions—proliferate, a critical bottleneck has emerged: they can't easily collaborate. Much like the early days of Wi-Fi or the web, today's agents operate in isolated silos, limiting their collective potential. Cisco's response, AGNTCY (pronounced "agency"), is now poised to become the glue that binds them. Donated to the Linux Foundation and backed by industry titans including Dell, Google Cloud, Oracle, and Red Hat, AGNTCY aims to establish an open, vendor-neutral 'Internet of Agents.' Without such a standard, the AI ecosystem risks fragmentation, where proprietary systems dominate or competing protocols stall progress—echoing historical standards wars that delayed innovations for decades.

The Interoperability Imperative

AI agents promise transformative efficiency, from automating business workflows to managing smart devices. Yet, as Vijoy Pandey, GM and SVP of Outshift by Cisco, underscores: "> Building the foundational infrastructure for the Internet of Agents requires community ownership, not vendor control. The missing piece isn't smarter agents—it's complete infrastructure that lets any agent work with any other agent, regardless of who built it or where it runs." This fragmentation isn't just inconvenient; it stifles innovation. Developers waste resources reinventing communication layers, while enterprises face integration nightmares. AGNTCY, first open-sourced by Cisco in March 2025 and now under Linux Foundation governance, provides a unified framework to prevent this—much like TCP/IP did for the internet.

Inside AGNTCY's Technical Framework

AGNTCY tackles four core challenges through modular, open-source components:
- Agent Discovery: The Open Agent Schema Framework (OASF) acts as a 'DNS for agents,' enabling them to locate and understand each other's capabilities dynamically.
- Agent Identity: A cryptographic system ensures verifiable identities, allowing secure, authorized actions across organizational boundaries.
- Agent Messaging: Secure Low-latency Interactive Messaging (SLIM) handles multi-modal communication (e.g., text, voice, data streams) with quantum-safe encryption and human-in-the-loop support.
- Agent Observability: Specialized SDKs monitor intricate multi-agent workflows, crucial for debugging probabilistic AI behaviors and ensuring reliability.

This architecture doesn't just replace emerging standards like Anthropic's Model Context Protocol (MCP) or the Agent2Agent (A2A) protocol—it integrates them. For instance, SLIM can transport A2A or MCP communications, while AGNTCY's directories help agents discover peers using these protocols. The observability tools then provide cross-protocol insights, turning potential chaos into a cohesive ecosystem.

Why Developers Should Care

For engineers, AGNTCY reduces the friction in building multi-agent systems. Instead of wrestling with custom integrations, developers can focus on agent logic, leveraging AGNTCY's GitHub-hosted codebase to ensure compatibility. John Roese, Dell's Global CTO, highlights the broader impact: "> Interoperability is central to Dell's agentic AI vision. Interworking technologies must accommodate agents wherever they are deployed, whether in public clouds, private data centers, the edge, or on devices." Early estimates suggest such standardization could generate $450 billion in value by 2028 by streamlining deployments.

Under the Linux Foundation's stewardship, AGNTCY invites community contributions to refine its infrastructure—a proven model for fostering adoption. As agentic AI evolves from novelty to necessity, this open framework could be the catalyst that finally lets autonomous systems achieve their collaborative potential, turning isolated tools into a symphony of intelligence.

Source: ZDNet, Steven Vaughan-Nichols