Build agents that can work with other agents.
Create AI agents as part of your AI System and connect them to other agents through the Agent2Agent (A2A) protocol.

Agents as part of the system
Agents run alongside business logic, APIs, MCP tools, workflows, and data within the same AVAP runtime.
They can use the capabilities of the system without requiring a separate agent backend.
Build agents around real capabilities
An agent can interact with the business logic and resources that already exist in the system.
This keeps agents connected to the actual system they operate in.
Give agents access to:
- Business capabilities
- APIs
- MCP tools
- Data and knowledge
- Workflows
- Other agents
Agent-to-Agent communication
Use A2A to allow agents to discover and communicate with capabilities provided by other agents.
Agents can collaborate across systems without exposing or duplicating the underlying implementation.
Govern agent interactions
Control what agents can access and what actions they can perform.
Permissions, policies, and system governance apply to agent interactions alongside other system capabilities.
Evolve agents continuously
Modify agent logic, tools, workflows, and integrations without rebuilding the surrounding application.
The same runtime supports development and execution.
Agents within the AI System
Agents are not isolated applications.
They are executable components of the system, connected to its logic, data, tools, and other agents.