OpenAI, Google and Anthropic Now Share Same Key Protocol
Google has decided to adopt Anthropic's Model Context Protocol (MCP) for its Gemini AI models, just weeks after OpenAI made a similar commitment.
At the same time, Google is launching its own Agent2Agent (A2A) protocol. For users, this means AI assistants will soon become much more capable - they'll be able to access your important information when needed and work together with other specialized AI tools to complete complex tasks for you. MCP helps AI access data, while A2A enables AI agents to collaborate.
Google Embraces Rival's Protocol
Google DeepMind CEO Demis Hassabis announced yesterday (Wednesday) that Google will add support for Anthropic's Model Context Protocol to its Gemini models and SDK (Software Development Kit, the tools developers use to build applications). This unexpected alliance between competitors underscores the growing importance of standardization in AI.
"MCP is a good protocol and it's rapidly becoming an open standard for the AI agentic era," Hassabis wrote on X. "Look forward to developing it further with the MCP team and others in the industry."
The adoption comes just weeks after OpenAI made a similar commitment, creating unprecedented alignment among the three leading AI research organizations.
Breaking Down the Barriers
MCP, which Anthropic recently open-sourced, addresses a fundamental limitation of current AI systems: their isolation from essential data sources. Even the most advanced models struggle when cut off from business tools, content repositories, and development environments where critical information resides.
The protocol enables two-way connections between AI models and data sources through "MCP servers" that expose information and "MCP clients" that connect to those servers. This standardized approach eliminates the need for fragmented, custom integrations for each data source.
Companies including Block, Apollo, Replit, Codeium, and Sourcegraph have already integrated MCP into their platforms, allowing AI agents to retrieve contextual information more effectively and produce more accurate, relevant responses.
A2A: The Next Frontier in AI Collaboration
While MCP focuses on AI-to-data connections, Google's newly announced Agent2Agent (A2A) protocol tackles a different challenge: enabling AI agents to communicate with each other across enterprise systems.
Google explicitly designed A2A to complement Anthropic's MCP. Together, they solve two critical problems: MCP helps AI access your data, while A2A lets different AI tools collaborate with each other. This combination means your AI assistants can both find the information they need and work together on complex tasks.
Launched with support from over 50 technology partners including Atlassian, Box, Cohere, Intuit, Salesforce, and SAP, A2A creates a standard framework for autonomous agents to collaborate across siloed data systems and applications—even when built by different vendors or using different frameworks.
This interoperability is crucial for enterprises deploying multiple specialized AI agents, allowing them to work together on complex tasks spanning different systems and departments.
How the Protocols Complement Each Other
The relationship between MCP and A2A highlights their complementary nature. MCP connects models to data sources, while A2A enables agents to communicate with each other. Google explicitly acknowledged this connection, noting that "A2A is an open protocol that complements Anthropic's Model Context Protocol."
This creates a complete ecosystem where:
AI models access data through MCP
Specialized agents built on these models perform specific tasks
These agents communicate and collaborate via A2A
Technical Underpinnings
A2A operates through several key mechanisms:
Capability discovery: Agents advertise their capabilities using "Agent Cards" in JSON format
Task management: Communication is organized around tasks with defined lifecycles
Collaboration: Agents exchange messages containing context, replies, and instructions
User experience negotiation: Messages include "parts" with specified content types
The protocol builds on existing standards including HTTP, SSE, and JSON-RPC, making it easier to integrate with existing enterprise IT infrastructure.
Real-World Applications
The combined power of these protocols enables complex workflows that were previously difficult to implement. For example, a hiring manager could use their primary agent to coordinate with specialized recruiting agents, interview scheduling systems, and background check services—all working together across different platforms and data sources.
This level of integration could dramatically reduce manual coordination and data transfer between systems, allowing enterprises to automate complex multi-step processes.
Industry Response
The rapid adoption of these protocols by major players signals a significant shift toward standardization in AI integration. Industry partners have expressed enthusiasm about the potential impact.
"We believe A2A will add significant value for customers, whose AI agents will now be able to work across their entire enterprise application estates," Google stated in its announcement.
Salesforce noted that it is "leading with A2A standard support to extend our open platform, enabling AI agents to work together seamlessly across Agentforce and other ecosystems."
Future Outlook
Both protocols are being developed as open standards with community contribution encouraged. Google plans to release a production-ready version of A2A later this year, while Anthropic continues to expand MCP's capabilities and integration options.
The convergence around these complementary standards suggests a potential future where AI systems can freely access information and collaborate with specialized agents across organizational boundaries—dramatically increasing their utility and impact.
For businesses using AI, this means it's getting smarter to choose AI systems that can work together rather than isolated tools. Companies should look at how these new standards could help connect their existing software and data.
Simply put, these protocols are a big step toward AI assistants that can actually deliver on their promise – helping you get real work done by accessing the information you need and coordinating with other tools to complete tasks.