Here’s the translation into American English:
With the backing of over 50 tech companies and global consulting firms, Google’s new open standard aims to lay the foundation for a true era of interoperability among intelligent agents.
Google has taken a crucial step towards the future of artificial intelligence with the introduction of Agent2Agent (A2A), an open protocol that will allow autonomous AI agents to communicate, collaborate, and coordinate tasks with each other, regardless of the technology or provider they were developed with.
The protocol is supported by a broad coalition of leading tech players and consulting firms, including names like SAP, Salesforce, MongoDB, Box, Atlassian, JetBrains, Accenture, Deloitte, BCG, and McKinsey, among others. They all share a common vision: to build a truly interoperable and decentralized AI ecosystem capable of tackling complex tasks in increasingly demanding business environments.
Why a protocol like A2A?
In the current context, intelligent agents are being deployed to automate and scale processes in nearly every sector: from customer service and hiring to supply chain planning. However, most of these agents work in isolation, limiting their collaborative potential.
With A2A, Google proposes a standard that will break those silos and unlock multi-agent workflows, even when agents have been developed with different technologies or belong to different providers. The key is to provide a common language for them to understand and work together.
Features of the A2A protocol
The design of A2A is based on five fundamental principles:
- True interoperability: Agents can collaborate without the need to share context, tools, or memory. A2A enables multi-agent use cases beyond the traditional “agent as a tool” approach.
- Based on existing standards: Utilizes widely adopted technologies such as HTTP, SSE, and JSON-RPC, making it easy to adopt in already established business environments.
- Security by default: Provides enterprise-level authentication and authorization, aligned with OpenAPI standards.
- Compatible with long-running tasks: A2A allows managing tasks that may extend over hours or days, providing real-time updates and constant synchronization between agents.
- Multimodality: Supports multiple input and output formats, including text, images, audio, or video, enabling richer experiences tailored to the end user.
How A2A works
In its basic structure, A2A connects a client agent—which defines a task—with a remote agent that executes it. From there, various functions come into play, such as:
- Capability discovery: Through an “agent card” in JSON format, agents announce their capabilities, enabling the client agent to identify the most suitable one for each task.
- Task lifecycle management: Each task has a status, can generate results (artifacts), and can be completed immediately or with progressive updates.
- User experience negotiation: Agents negotiate how to present the results based on the capabilities of the user’s environment (e.g., whether it supports web forms or videos).
A practical example: automated hiring
One of the most representative use cases is the automation of hiring processes. In a system implementing A2A, an agent can be tasked with finding suitable candidates for a position. This agent communicates with others specialized in talent sourcing, and once the profiles are identified, another agent can schedule interviews, and yet another can carry out background checks. The entire process occurs through a chain of collaboration among interoperable agents, with no direct human intervention.
A strategic and technological commitment
A2A is not just a technical proposal, but also a strategic commitment to a more open, efficient, and collaborative future for AI. The partners who participated in its definition emphasize that this protocol is a key piece to address the current challenges of scalability, fragmentation, and integration in enterprise artificial intelligence.
From SAP to LangChain, including MongoDB, JetBrains, Elastic, and PayPal, tech leaders agree that A2A establishes the foundations for a new paradigm where agents can interact safely, efficiently, and flexibly, regardless of their origin.
This move by Google—supported by broad industry backing—anticipates a profound transformation in how intelligent systems are designed, connected, and operated. Interoperability is no longer an aspiration: it is beginning to become a reality.