Linux foundation Announced on Open source summit In Denver that it will host now Agent 2 Agent (A2A) ProtocolInitially developed by Google and now supported by more than 100 leading technology companies, A2A AI is an important new open standard for secure and inter -conversion among agents.
Also: What are AI agents? How to use a team of individual assistants
In its main presentation, Google Staff Software Engineer, Mike Smith told the conference that the A2A protocol has been developed to make it easier to add custom extensions to the main specification. Additionally, the A2A community is working on making AI agents easy to assign unique, which is improving governance and security.
The A2A protocol is designed to solve one of the most pressured challenges of AI: enabling autonomous agents-Software institutions capable of independent action and decision-making to search for a each other, exchange information safely, and cooperate in uneven platforms, sellers and structures. Under the hood, A2A makes it one by making it Agent card,
Too: Amazon’s Andy Jassi says AI will do some work but will make others more interesting ‘
An agent card is a JavaScript object notation (JSON) metadata document that describes its purpose and provides instructions to access it through the web URL. The A2A is widely adopted widely adopted by web standards adopted, such as HTTP, JSON-RPC and Server-Sent events (SSEs) to ensure widespread compatibility and ease of integration. By providing a standardized, seller-plate communication layer, the A2A breaks the silos that historically limit the potential of the multi-agent system.
For safety, A2A comes with enterprise-grade certification and authority, including support Json web tokens (jwts), Openid Connect (OIDC)And Transport layer securityThis approach ensures that only authorized agents can only participate in the workflows, protect sensitive data and agents identity. While the security foundation is applicable, the developers of the conference admitted that they would have a difficult slogan to integrate them, especially certified agents.
A new era of productivity
So, what does it mean to adopt A2A for IT professionals? In an interview, Amazon Web Services (AWS) explained, “Amazon Web Services (AWS) Principal Advocate of Amazon Web Services (AWS) explained,” Say, “You want to book a train ride for Copenhagen, then a hotel, and perhaps for a fancy restaurant, you have a fancy restaurant, your inputs and personal tasks, and a 2A in the conversation, and A2A, and A2A in the A2A Connects agents. Other, stop the task, and finally brings a response to the final user. ,
As the Linux Foundation Executive Director Jim Zemlin said in his main speech: “By joining the Linux Foundation, A2A is ensuring long-term neutral neutrality, cooperation and governance that will unlock the next era of agent-to-agent operating productivity.” Zemlin hopes that the A2A will become a foundation stone for the construction of the Multi-agent AI system.
AI companies agree. A coalition of industry giants supports the A2A project. Apart from Google and AWS, Cisco, Microsoft, Slesforce, SAP, and Servicenow, the protocol back.
Also: 5 ways you can widen the AI skill difference in your business
The A2A is expected to work hand-in-glav with anthropic model reference protocol (MCP). While both are open standards that enable strong and interofforable AI agents ecosystem, they have different objectives.
MCP offers a standardized way to access external equipment, application programming interface (API), and data sources for an individual AI agent. It acts as a universal interface, “USB-C port for AI applications“Enabling AI agents to enhance their abilities by joining services such as search engines, databases, or third-party APIs.
On the other hand, the A2A focuses on enabling communication, cooperation and work delegation between several AI agents, regardless of its seller, outline, or underlying technology.
Also: AI will not take your job, but it will definitely happen
Classy explained as a blog by AI Company, “This Two protocols solve different parts of communication problem: A2A focuses on how agents communicate with each other (horizontally), while MCP focuses on how a single agent connects to the tool or memory (vertical). ,
Protocol
Industry experts also hope that MCP will be used in vertical integration. For example, agents will use MCP to collect or process the information required for their tasks, such as rebuilding user data from the upstream customer relationship management program or triggering cloud service.
With A2A horizontal integration, when a task requires several agents with special expertise, the A2A allows these agents to search each other, share references, hand over, sub -cups and coordinate to solve complex problems.
In practice, as AI expert Aishera said, when A This support agent, using A2A, receives a ticket about the slow application Performance, it uses MCP to collect clinical data from monitoring equipment. When a problem extends to several domains, such as network, database, or application, the agent can use A2A to hand over parts of other special agents to parts of the investigation, each of which can use MCP to reach its relevant equipment.
Also: 4 ways to convert AI into their business gains
But, the good as the A2A is, Barth warns, “We are still in the early stages from the perspective of a protocol and standardization. And I don’t think we are yet to call; There is just a protocol to deal with matters of using all the use. MCP is initially to add models with devices, and there is a huge community around it.
Working at the same place where A2A and MCP start and end will be challenging. Apart from A2A, there is another agent-to-agent protocol, Agent communication protocolThis open-source protocol is based on the popular Representative State Transfer (REST) Software Architecture from IBM, and also supports the Linux Foundation.
In short, let the best protocol win. However, which is already certain that all this effort will pay dividends in the long term. As the berth concluded, “Gartner says that one third of all applications and enterprises will be operated by agent AI by 2028Therefore, the developers have never had a better time to start construction, start learning. ,
