A2A - Google's first open-source standard intelligent agent interaction protocol

What is A2A?

A2A (Agent2Agent Protocol) is Google's first open-source standard intelligent agent interaction protocol, enabling AI agents built on different frameworks and from different vendors to collaborate. Based on standardized communication methods, A2A breaks down system silos, allowing agents to securely exchange information and coordinate actions. A2A supports multiple modalities (such as text, audio, and video), long-duration task processing, and real-time feedback. Built on existing standards (such as HTTP and JSON-RPC), it is easily integrated with existing IT systems. The initial batch of participating companies includes over 50, including Atlassian, Box, and MongoDB.

The main functions of A2A

• Promote agent collaborationA2A enables seamless collaboration between AI agents built on different frameworks and by different vendors.
• Supports multiple modalitiesA2A supports multiple interaction methods such as text, audio, and video to meet the needs of different scenarios.
• Task management and status updateA2A defines a "task" object with a lifecycle (e.g., submitted, in progress, input required, completed, failed, canceled). For long-running tasks, such as complex data analysis or lengthy file processing, A2A provides real-time feedback to the user on task progress and status updates.
• User experience negotiationA2A enables agents to negotiate user experience and adjust interaction methods based on user devices and preferences.
• Security and AuthenticationA2A supports enterprise-level authentication and authorization, ensuring the security and compliance of data exchange.

A2A design principles

• Leveraging agency capabilitiesSupports agents to collaborate in a natural, unstructured manner.
• Based on existing standardsBased on existing standards such as HTTP, SSE, and JSON-RPC, it is easy to integrate with existing IT systems.
• Default securitySupports enterprise-level authentication and authorization, consistent with OpenAPI's authentication scheme.
• Support long-term tasksIt supports tasks ranging from quick ones to in-depth research tasks that may take hours or even days, providing real-time feedback and status updates.
• Modality independentIt supports multiple interaction methods such as text, audio, and video.

How A2A works

• Capability DiscoveryThe agent advertises its capabilities based on "Agent Card" (JSON format), and the client agent can identify and select the most suitable agent to perform the task.
• Task ManagementCommunication between the client and the remote agent is task-completion oriented. Task objects have a lifecycle and can be completed immediately or run for a long time.
• cooperationAgents can send and receive messages, including context, replies, task outputs, or user commands.
• User experience negotiationEach message contains a "part," which is a complete content fragment. Each part has a specified content type, and the client and remote agent can negotiate the correct format.

A2A Technical Principles

• Agent CardAgent Card: An agent's "business card," storing its capabilities, skills, endpoint URLs, and authentication requirements in JSON format. Clients obtain the Agent Card by accessing /.well-known/agent.json, and then discover and select suitable agents for collaboration.
• A2A Server and A2A ClientThe A2A Server is an HTTP endpoint that implements the A2A protocol methods, responsible for receiving requests and managing task execution; the A2A Client is a client or another intelligent agent that sends requests (such as tasks/send). Both communicate based on the HTTP protocol.
• Tasks and MessagesA task is the core unit of work in A2A. The client initiates a task by sending a message (containing a task description). The message consists of multiple "parts" and may contain text, files, or structured data. During execution, the task goes through different states, which are fed back to the client in the form of messages.
• Streaming and push notificationsFor long-running tasks, A2A supports Server-Sent Events (SSE), allowing clients to receive real-time task progress updates. A2A also supports push notifications, where the server proactively sends task updates based on the webhook URL provided by the client.
• Based on existing standardsThe A2A protocol is built on existing popular standards (such as HTTP, SSE, and JSON-RPC) and supports integration into existing IT architectures, lowering the barrier to adoption for enterprises.

A2A Partners

A2A has received support from more than 50 technology partners and service providers, including Atlassian, Articul8, Arize AI, Box, C3 AI, BCG, Capgemini, and Cognizant.

A2A project address

• Project official website:https://developers.googleblog.com/en/a2a
• GitHub repository:https://github.com/google/A2A

A2A application scenarios

• Enterprise Process AutomationIt connects systems from different departments to automatically complete tasks such as order processing and inventory management.
• Cross-platform customer serviceIntegrating intelligent customer service across multiple platforms to provide more comprehensive and timely customer support.
• Recruitment process optimizationIt can coordinate with multiple recruitment platforms and internal systems to quickly screen candidates and arrange interviews.
• Supply chain collaborationReal-time sharing of information across all stages of the supply chain optimizes inventory and logistics.
• Smart OfficeIntegrate intelligent agents from office tools to automatically schedule meetings and remind tasks, thereby improving collaboration efficiency.