Collaborate with AI Agents Across Platforms using Agent2Agent (A2A) Protocol

The published AI agents in Oracle Fusion Cloud Applications can collaborate with AI agents in other platforms using the A2A protocol. A2A is an open, industry-backed protocol that enables AI agents to discover, understand, and call other agents in a standardized way. In AI Agent Studio, A2A serves as the foundation for seamless cross-agent collaboration across Fusion Applications and external ecosystems, eliminating the need for custom point-to-point integrations.

Here are the key features of A2A:

A standardized way to describe agents through an Agent Card
A standard request-and-response model for agent interactions
A shared lifecycle framework for managing long‑running and asynchronous tasks

Here's an example scenario for using A2A. If your organization uses an Oracle AI agent for sales managers, that agent can reach out to a third-party agent to get travel options and cost quotes when considering the sales calls. Similarly, third-party platforms can call Oracle AI agents using A2A. Another example scenario for using A2A is when your organization has a general purpose AI chat assistant for employees. You can reach out to subject matter expert agents in Oracle Fusion Applications when customers ask about HR Sickness and Absence Policy or Vacation Bookings. Here's how the interaction using A2A works:

Each agent publishes an Agent Card that describes the agent to the third-party platform
Third-party platforms can then call the Fusion Applications agent in a consistent and secure way

The Fusion Applications A2A API provides clients a practical way to discover, call, poll, and stream workflow-backed agents using A2A-style endpoints and method names. As an authenticated, client-facing API, it offers the following key properties:

A workflow-backed agent model
A task-oriented asynchronous interaction model
A pragmatic subset of broader A2A capabilities