Most enterprise AI deployments today are a collection of smart silos. You have an agent that summarizes documents. Another that queries your CRM. A third that handles scheduling. Each one is useful, but none of them know the others exist. Getting them to collaborate means writing custom integration code that breaks every time one of the underlying models or platforms changes.
Google’s Agent2Agent (A2A) protocol, introduced in April 2025 and handed to the Linux Foundation in June 2025, will solve this problem at the standards level. Think of it as HTTP for AI agents; a common language that lets agents discover each other, authenticate, delegate tasks, share context, and report progress regardless of the framework on which they’re built or on what cloud they run.
How It Actually Works
A2A is built on three core primitives. Agent Cards are JSON metadata that describe what an agent does, what inputs it needs, and how to authenticate with it. Any other agent can read this manifest and determine whether this agent has what it needs. Task Management gives agents a standardized way to delegate work and receive real-time status updates. Long-running jobs that span hours or days are supported natively. Context Sharing lets agents pass relevant information between each other. Internal data the receiving agent doesn’t need stays hidden.
The protocol runs over HTTP/HTTPS with JSON-RPC, supports OAuth2, API keys, and mTLS for authentication. It’s designed to slot into existing enterprise infrastructure rather than require a new runtime. The security model is explicit. OAuth 2.0, digital signatures, and role-based access control are part of the spec, not an afterthought. For regulated industries, that auditability is really going to matter.
A2A also supports modalities beyond text. Audio and video streaming are part of the protocol. That makes it applicable to voice agents, real-time translation tools, and multimedia workflows a text-only protocol couldn’t handle. For ISVs building in those spaces, that’s a meaningful capability that most competing approaches don’t offer.
Crucially, A2A complements Anthropic’s Model Context Protocol (MCP) rather than competing with it. MCP handles vertical integration; connecting an agent to a specific tool or data source. A2A handles horizontal integration; connecting agents to other agents. Most serious multi-agent architectures will use both, and the two protocols are designed to work together.
The Ecosystem Behind It
Protocol adoption lives or dies on who builds to it. At launch in April 2025, A2A had 50+ technology partners. By July 2025, that number had grown to over 150 organizations! The list includes names that matter in enterprise software. Atlassian, Box, Cohere, Intuit, LangChain, MongoDB, PayPal, Salesforce, SAP, ServiceNow, and Workday all supported the launch. Adobe and ServiceNow are already using A2A to make their agents interoperable within enterprise environments.
That adoption curve matters for ISVs evaluating whether to build to A2A. A protocol with 150+ enterprise backers and Linux Foundation governance is a safe bet in a way that a proprietary vendor standard is not. If your customers are assembling agent ecosystems that include Salesforce, SAP, and ServiceNow, building to A2A means your agents interoperate out of the box. No custom integration work required on the customer’s side.
What This Means for ISVs
For ISVs building on Google Cloud, A2A changes the architecture of what you can ship. Internally, if you run multiple specialized agents across your own platform, A2A gives them a standard coordination layer. You stop writing bespoke glue code between them and start building a composable agent mesh that grows without every new addition requiring integration work.
The product angle is more interesting. If your software product includes AI capabilities, A2A means your agents can interoperate with your customers’ existing agent ecosystems and with third-party agents from partners already building to the spec. Your product stops being an island and becomes a participant in the broader agent network your customers are assembling. Additionally, Google Cloud’s AI Agent Marketplace gives those agents a distribution channel, letting customers discover and deploy A2A-compatible agents directly through the marketplace.
That is a real differentiator in enterprise sales. Procurement teams buying AI-powered software increasingly ask how it fits into their wider automation strategy. A product built on A2A has a concrete, standards-based answer. One built on proprietary agent integration doesn’t.
The Competitive Angle
Microsoft has its own agent interoperability work inside the Copilot ecosystem, but it’s largely proprietary and optimized for Microsoft-to-Microsoft integration. AWS has multi-agent orchestration in Bedrock, but no open cross-vendor protocol. A2A under the Linux Foundation is the only vendor-neutral open standard. For ISVs who sell across cloud environments and need a story that works with any customer’s stack, that neutrality matters.