#agent2agentprotocol

Announcement of A2A Protocol.

New Agent Interoperability period

AI agents can boost productivity by handling many repetitive or challenging everyday tasks. Businesses are designing and using autonomous agents to scale, automate, and optimise workplace processes including customer service, supply chain planning, and laptop purchases.

To maximise agentic AI benefits, these agents must collaborate in a dynamic, multi-agent ecosystem across fragmented data sources and applications. Allowing agents to communicate, even if they were built in different frameworks or by different suppliers, reduces long-term costs, boosts autonomy, and boosts productivity.

More than 50 technology partners, including Atlassian, Box, Cohere, Intuit, Langchain, MongoDB, PayPal, Salesforce, SAP, ServiceNow, UKG, and Workday, and top service providers, including Accenture, BCG, Capgemini, Cognisant, Deloitte, HCLTech, Infosys, KPMG, McKinsey, PwC, TCS, and Wipro, are launching a new, open protocol called Agent2Agent today.

AI agents may coordinate activities, securely share information, and interact via the A2A protocol on corporate platforms or applications. Since clients' AI agents may now run throughout their corporate application estate, the A2A framework should benefit them.

This united effort envisions a period where AI agents can seamlessly automate complex corporate processes and unleash unprecedented creativity and efficiency, regardless of technology.

Open protocol A2A enhances Anthropic's Model Context Protocol (MCP), which offers agents tools and context. Google used their expertise scaling agentic systems to design the A2A protocol to solve customer problems while implementing large-scale, multi-agent systems.

A2A lets consumers combine agents from multiple sources and lets developers design agents that can communicate with other protocol agents. Most significantly, standardised agent management across cloud environments and platforms benefits enterprises. They believe universal interoperability is essential for collaborative AI agents to succeed.

Partners in A2A Protocol

A2A is an open protocol that lets agents interact together regardless of framework or vendor. When building the procedure with others, it followed five key guidelines:

Accept agentic capabilities: A2A allows agents to collaborate in their natural, unstructured ways without context, memory, or tools. We enable actual multi-agent scenarios without limiting agents to tools.

Build on existing standards: The protocol is built on HTTP, SSE, and JSON-RPC, making it easier to incorporate into enterprises' IT stacks.

Secure by default: A2A's authentication schemes match OpenAPI's at launch and handle enterprise-grade authorisation and authentication.

Support for long work: Everyone created A2A to be adaptive and perform effectively in anything from quick assignments to in-depth research that may take hours or days with individuals. A2A can provide real-time feedback, notifications, and state changes during this process. Agnostic about modality We created A2A to support streaming audio and video since the agentic world isn't just text.

How A2A works

A2A simplifies “client” and “remote” agent communication. The client agent creates and transmits tasks, while the remote agent executes them to supply the correct information or take the necessary action. This interaction requires several skills:

Ability discovery: Agents can advertise their capabilities via a JSON “Agent Card”. An agent can identify the best candidate for a position and utilise A2A to connect with the distant agent.

Remote agents and clients communicate to execute tasks and meet end-user needs. The protocol defines a lifecycled “task” object. It can be completed immediately or each agent can update each other on a long-term project. A task produces a “artefact”.

Secure Collaboration: Agents can convey context, answers, artefacts, and user instructions. User experience negotiation: Every communication has "parts," such as a created picture. Because each component has a content type, the client and remote agents may explicitly negotiate the user's UI capabilities, such as iframes, video, web forms, and more, and the requisite format.

Practical example: candidate sourcing

A2A cooperation simplifies software engineer recruiting. A recruiting manager may assign an agent to discover people that match a job description, location, and skill set in Agents pace. The agent then contacts other speciality agencies to identify applications.

The customer can ask their agent to schedule more interviews after obtaining these recommendations, speeding up candidate selection. After the interview, another agent may do background checks. This is only one example of how AI agents must collaborate across systems to select a good candidate.

The future of agent interoperability