Google Cloud officially announced the launch of its new open-source Agent2Agent (A2A) protocol, designed to break down the barriers between current artificial intelligence (AI) agents and enable seamless collaboration and information exchange between AI agents built by different vendors and using different frameworks. This move marks a step towards a more open, efficient, and intelligent new era in the AI field.
Breaking Down Barriers: Enabling Free "Conversation" Between Agents
As enterprises increasingly deploy autonomous AI agents, enabling these agents to collaborate in complex enterprise environments to handle routine or complex cross-system, cross-application tasks has become crucial for boosting productivity. However, due to the lack of unified standards, different agents often struggle to communicate and collaborate effectively, creating "information silos."
Google's newly released A2A protocol addresses this pain point. As an open protocol, A2A provides a standardized approach to communication, secure information exchange, and collaborative actions between AI agents. Regardless of the vendor or underlying technology used, as long as the agents adhere to the A2A protocol, true multi-agent scenarios can be realized, breaking free from the traditional limitation of confining agents to single "tools."
Five Core Design Principles: Building a Strong and Reliable Interoperability Foundation
The A2A protocol was designed with five key principles to ensure its adaptability to future complex and evolving AI application scenarios:
- Embrace agentic capabilities: A2A focuses on enabling agents to collaborate in their natural, unstructured ways, working efficiently even without shared memory, tools, and context.
- Build on existing standards: The protocol is built upon popular existing standards such as HTTP, SSE, and JSON-RPC, significantly reducing the difficulty for enterprises to integrate it into their existing IT infrastructure.
- Secure by default: A2A inherently supports enterprise-grade authentication and authorization, with security comparable to OpenAPI authentication schemes.
- Support for long-running tasks: A2A is highly flexible and supports various scenarios, from quick tasks to in-depth research requiring hours or even days to complete. Throughout the process, A2A provides users with real-time feedback, notifications, and status updates.
- Modality agnostic: Recognizing that the intelligent world extends beyond text, A2A also supports multiple modalities, including audio and video streams.
Core Functionality Analysis: How Agents "Communicate" and "Collaborate"?
The A2A protocol defines key capabilities for agent interaction:
- Capability discovery: Agents can advertise their capabilities through JSON-formatted "Agent Cards," allowing client agents to find the most suitable remote agents for specific tasks.
- Task management: Communication between agents is task-oriented. The protocol defines "task" objects and their lifecycle; tasks can be completed immediately or tracked through synchronous communication between agents for long-running tasks. The output of a task is called an "artifact."
- Collaboration: Agents can send messages to each other to convey context, replies, artifacts, or user instructions.
- User experience negotiation: Each message contains "parts," a complete data block of a specific content type (e.g., a generated image). Client and remote agents can negotiate the correct format required for user interface capabilities, such as iframes, videos, and web forms.
It's worth noting that the release of the A2A protocol has received the support and contributions of over 50 technology partners and leading service providers. These include tech giants such as Atlassian, Box, Cohere, Intuit, Langchain, MongoDB, PayPal, Salesforce, SAP, ServiceNow, UKG, and Workday, as well as well-known consulting firms like Accenture, BCG, Capgemini, Deloitte, KPMG, and PwC. Numerous partners have stated that the A2A protocol will help drive broader adoption of AI in enterprises, enabling more efficient and intelligent workflows.
Open Source Empowerment: Building a Thriving Agent Ecosystem
Google emphasizes that the A2A protocol will be released as open source, encouraging community participation in the protocol's evolution, including submitting ideas, contributing documentation, and participating in community discussions. Currently, a complete draft specification and code examples are available on the official A2A website. Google plans to release a production-ready version of the protocol later this year.
Google believes that the A2A protocol will usher in a new era of agent interoperability, fostering innovation and creating more powerful and versatile agent systems. By breaking down the barriers between agents, A2A is expected to unlock significant potential in areas such as enterprise automation, customer service, and supply chain management.
