Google Cloud has launched Agent2Agent (A2A), an open protocol enabling AI agents from different vendors and frameworks to communicate, exchange information, and coordinate actions across enterprise platforms.
Over 50 technology partners including Atlassian, Salesforce, SAP, and MongoDB, along with leading service providers like Accenture, Deloitte, and KPMG, are supporting and contributing to the A2A protocol.
The protocol is designed with five key principles: embracing agentic capabilities, building on existing standards, security by default, support for long-running tasks, and modality agnosticism.
A2A facilitates communication between "client" and "remote" agents through capability discovery, task management, collaboration, and user experience negotiation.
A production-ready version of the protocol is expected to launch later in 2025, with the current specification draft and code samples available on the A2A website.
The Vision and Purpose of A2A Protocol
A2A aims to enable AI agents to collaborate in a dynamic, multi-agent ecosystem across siloed data systems and applications, increasing autonomy and productivity while lowering long-term costs.
The protocol represents a shared vision of a future where AI agents can seamlessly collaborate regardless of underlying technologies.
A2A complements Anthropic's Model Context Protocol (MCP), which provides tools and context to agents.
Google designed A2A based on internal expertise in scaling agentic systems and to address challenges identified in deploying large-scale, multi-agent systems.
The protocol allows developers to build agents capable of connecting with any other agent built using the protocol.
Users gain flexibility to combine agents from various providers across diverse platforms and cloud environments.
Core Design Principles of the A2A Protocol
Embrace agentic capabilities: Enables agents to collaborate in natural, unstructured modalities even without shared memory, tools, and context, supporting true multi-agent scenarios.
Build on existing standards: Leverages popular standards including HTTP, SSE, and JSON-RPC for easier integration with existing IT stacks.
Secure by default: Designed to support enterprise-grade authentication and authorization, with parity to OpenAPI's authentication schemes.
Support for long-running tasks: Flexible design supporting quick tasks to deep research that may take hours or days, with real-time feedback and state updates.
Modality agnostic: Supports various communication modalities beyond text, including audio and video streaming.
Technical Framework and Functionality
A2A facilitates communication between a "client" agent (formulates and communicates tasks) and a "remote" agent (acts on tasks to provide information or take action).
Capability discovery: Agents advertise capabilities using an "Agent Card" in JSON format, allowing client agents to identify the best agent for a task.
Task management: Communication is oriented toward task completion to fulfill end-user requests, with tasks having defined lifecycles and outputs known as "artifacts."
Collaboration: Agents send messages to communicate context, replies, artifacts, or user instructions.
User experience negotiation: Messages include "parts" (fully formed pieces of content) with specified content types, allowing agents to negotiate correct formats and UI capabilities.
The complete technical specification is available as a draft on the A2A website.
Real-World Application Example
The protocol can simplify hiring processes through agent collaboration:
A hiring manager can task their agent to find candidates matching specific job requirements.
This agent interacts with specialized agents to source potential candidates.
The user receives suggestions and can direct their agent to schedule interviews.
After interviews, another agent can facilitate background checks.
This example demonstrates how AI agents need to collaborate across systems to complete complex workflows.
Industry Support and Ecosystem Development
More than 50 technology partners are supporting and contributing to the A2A protocol, including:
Leading service providers: Accenture, BCG, Capgemini, Cognizant, Deloitte, HCLTech, Infosys, KPMG, McKinsey, PwC, TCS, Wipro
The protocol is being released as open source with clear pathways for contribution.
A production-ready version is planned for release later in 2025.
Partners see A2A as critical for enabling complex agent workflows, accelerating partner integrations, and moving the industry forward.
Partner Perspectives on A2A's Value and Impact
Technology & Platform Partners:
Atlassian: Views A2A as enabling richer forms of delegation and collaboration at scale
Articul8: Positions A2A as a first-class citizen in their "Agent-of-Agents" ModelMesh
Cohere: Emphasizes secure collaboration even in air-gapped environments
Intuit: Believes A2A will enable complex agent workflows and accelerate partner integrations
PayPal: Sees A2A as creating new ways for developers and merchants to create next-generation commerce experiences
SAP: Views A2A as pivotal for enabling their Joule AI to work across enterprise platforms
Salesforce: Focuses on extending their open platform to enable AI agents to work together across ecosystems
Services Partners:
Accenture: Describes A2A as "the bridge that will unite domain specific agents across diverse platforms"
Deloitte: Views agent-to-agent interoperability as foundational for agentic AI architecture evolution
KPMG: Considers A2A as providing the essential standard for different AI agents to collaborate effectively
TCS: Describes A2A as "the foundation for the next era of agentic automation"
Wipro: States that "open protocols like A2A will be the foundation of an ecosystem where AI agents drive innovation at scale"
Future Implications and Development Roadmap
A2A has potential to unlock a new era of agent interoperability, fostering innovation and creating more powerful agentic systems.
Google Cloud is committed to building the protocol collaboratively with partners and the community.
The protocol is being released as open source with clear pathways for contribution.
Full specification draft, code samples, and example scenarios are available on the A2A website.
A production-ready version of the protocol is planned for release later in 2025.
A2A represents a significant step toward AI agents that can seamlessly collaborate to solve complex problems across different systems and applications.
