A new industry standard for agentic workflows, dubbed Agntcy, has been accepted by the Linux Foundation, with a consortium of major technology companies, including Cisco, Google Cloud, Dell Technologies, Oracle, and Red Hat, joining the effort. The move signals a concerted push to accelerate the development of interoperable, large-scale agentic AI systems, or “internet of agents.”
Agntcy, first introduced in March by a group including Cisco’s Outshift incubator, Galileo, and LangChain, now stands alongside Google’s Agent2Agent (A2A) protocol as a new Linux Foundation project. This collaboration is designed to create a comprehensive framework for the discovery, composition, deployment, and evaluation of multi-agent systems.
“Just as the internet enabled the composition and deployment of servers, and the cloud did the same for services, we want to create a similar ecosystem for agentic endpoints,” said Vijoy Pandey, Senior Vice President and General Manager of Cisco’s Outshift.
Agentic AI, an emerging field, involves networks of autonomous software agents powered by large language models that collaborate to execute complex tasks. While major cloud providers and vendors like ServiceNow and IBM offer AI agents and orchestrators, large-scale workflows that span different frameworks and networks remain in early development. The Agntcy project is designed to fill this gap, proposing a collection of protocols to support more sophisticated and high-scale workflows.
Interoperability and Advanced Features
Agntcy is designed to be interoperable with other emerging standards, including A2A, which focuses on agent-to-agent communication, and Anthropic’s Model Context Protocol (MCP), which connects agents with data and tools. The key differentiator for Agntcy is its comprehensive suite of protocols, which includes Secure Low-Latency Interactive Messaging (SLIM). SLIM supports various communication patterns and incorporates built-in security features such as end-to-end encryption and authentication, addressing critical challenges in network security.
A Cisco spokesperson highlighted the project’s strategic alignment, stating, “The Agntcy project makes A2A agents and MCP servers discoverable, transparent through observability SDKs, and capable of exchanging messages efficiently and securely at scale through SLIM.” Cisco has representatives on both the Agntcy and A2A technical steering committees to ensure deep alignment between the two protocols.
Addressing Industry Needs
The need for a robust standard is particularly acute for companies developing enterprise-level AI applications. Ian Beaver, Chief Data Scientist at Verint, a contact center-as-a-service provider, noted that Agntcy’s focus on identity and access management makes it a more promising solution than existing protocols.
“Our use cases are very segmented into different roles and responsibilities… what one agent should be able to see is different from what another agent should see,” Beaver said. “MCP was largely insufficient because we can’t just expose all our data to an agent without proper identity and permissions.”
Beaver envisions Agntcy as a key component of Verint’s Intelligent Virtual Agent (IVA) Studio, enabling customers to easily publish and connect their bots to a broader network, thereby creating their own automated workforces.
Future Outlook and Collaboration
While industry analysts acknowledge the “wild west” nature of the current agentic workflow landscape, the strong backing for Agntcy from prominent tech firms under the Linux Foundation is a positive sign. Shamus McGillicuddy, an analyst at Enterprise Management Associates, believes Agntcy’s focus on creating directories and deeper networking has the potential to foster a true “internet of agents.”
Rob Strechay, an analyst at TheCube Research, echoed this sentiment, suggesting that having both Agntcy and A2A under the Linux Foundation could eventually lead to consolidation. “I hope that the Linux Foundation, based on contributions, will pick a winner,” Strechay said. “Minimizing network latency in these cases will be the name of the game and is in many AI deployments now.”
The collaboration between these tech giants under a neutral foundation suggests a shared vision for standardizing the nascent field of agentic AI, paving the way for scalable and secure multi-agent systems.
