AGNTCY Aims to Build the “Internet of Agents” Through Interoperability
One of the major goals for the future of AI is the ability for AI agents from different organizations to interact seamlessly. This vision, however, requires true interoperability. These agents are often built with different large language models (LLMs), data frameworks, and code, making communication a significant challenge. To overcome this, developers of these agents need a shared language for communication.
To address this hurdle, a consortium of companies, including Cisco, LangChain, LlamaIndex, Galileo, and Glean, has established AGNTCY. This open-source collective aims to create an industry-standard agent interoperability language, simplifying how any AI agent exchanges data with others.
Uniting AI Agents
“Just like when the cloud and the internet came about and accelerated applications and all social interactions at a global scale, we want to build the Internet of Agents that accelerate all of human work at a global scale,” said Vijoy Pandey, head of Outshift by Cisco, Cisco’s incubation arm, in an interview with VentureBeat. Pandey likened the project to the advent of the Transmission Control Protocol/Internet Protocol (TCP/IP) and the Domain Name System (DNS), which created the framework for the internet and allowed different computer systems to connect.
“The way we are thinking about this problem is that the original internet allowed for humans, servers, and web farms to all come together,” he said. “This is the Internet of Agents, and the only way to do that is to make it open and interoperable.”
Cisco, LangChain, and Galileo will take on primary maintainer roles within AGNTCY, with Glean and LlamaIndex contributing to the project. The collective anticipates adding further members as it develops.
Standardizing a Rapidly Evolving Field
AI agents cannot operate in isolation. To realize their full potential, they must be able to communicate with agents outside of an organization’s network. Interoperability is vital to this process. Setting standards is challenging in any industry, but it’s especially difficult in technology fields like AI, where models and upgrades change rapidly.
Several standardization efforts are underway in the generative AI space. LangChain, a core AGNTCY member, has already developed its Agent Protocol. Launched in November 2023, it allows LangChain agents to communicate with agents developed using AutoGen, CrewAI, and other frameworks. Anthropic announced its Model Context Protocol (MCP) at the same time to standardize how AI models access data sources. While MCP has gained some traction, it is not yet a widely accepted standard.
Yash Sheth, cofounder of AI evaluation platform Galileo, stresses the importance of standardization. “Standardization is needed; in fact, it will drive increased velocity for agentic adoption. Today, teams are building in silos, having to figure out how to develop their own infrastructure components from scratch,” Sheth said in an email. “Standardization of multi-agentic systems can only happen if these agents powered by non-deterministic models have a strong anchor in measuring and reporting their performance, accuracy, and reliability.”
Sheth acknowledged the complexity of enabling AI agent interoperability. AGNTCY “wants to encourage developers to extend these specs, APIs, and tools to suit their needs instead of reinventing the wheel, which will be crucial to achieving standardization.”
LangChain CEO Harrison Chase believes building a standard is achievable, especially given the increasing ease of developing the agents themselves. “Building agents is already possible, and being done. Replit, Klarna, LinkedIn, Uber, Appfolio and many others have all already done this. Agents aren’t a thing of the future; they are now. Now that we know how to build agents, the next step is to allow them to connect to each other. That is what a standard agent protocol will help enable,” Chase said.
A Unified Platform and Language
Pandey sees AGNTCY as more than a collection of codes. He envisions a platform on which customers can discover agents from various developers that align with the AGNTCY standard. Customers will be able to “stitch together all these agents on the AGNTCY platform so they can discover, compose, deploy, and evaluate as they build their workflows,” Pandey explained.
For AGNTCY to become a true industry standard, it needs widespread adoption. The team is actively seeking input from a broad range of industry players to ensure the platform meets diverse needs. This process will take time.
Even as AGNTCY develops, enterprises continue to experiment with and deploy AI agents. The future of AI may well see these agents seamlessly communicating, thanks to projects like AGNTCY.
