Chapter 14

The Social Brain of Agents: Inter-Agent Communication (A2A)

Interactive Graph (beta)

Toggle graph

Human intelligence is amplified through social exchange; agent intelligence scales through Inter‑Agent Communication (A2A). A2A is the protocol for meaning exchange, delegation, and collective intelligence across agents.

Neuroscience Analogy

• Neurons: Specialized units; intelligence from interaction.
• Humans: Societies achieve what individuals cannot.
• Agents: Specialists (retrieval, reasoning, planning, visualization) coordinate via structured messages.

Core Concepts

1. Agent Card = identity and capabilities.
2. Discovery = find collaborators via registries/URIs.
3. Communication = messages with attributes and parts; task lifecycle (submitted → working → completed).
4. Interaction Modes = sync, async polling, streaming (SSE), push (webhooks).
5. Security = authentication, encryption, audit.

A2A vs MCP

• MCP: One agent ↔ tools (sensory/motor interfaces).
• A2A: Many agents ↔ each other (social exchange).

Applications

Multi‑framework collaboration, automated workflow orchestration, dynamic information retrieval (scout ↔ analyst ↔ executor), and enterprise data coordination.

Conclusion

A2A turns isolated agents into a cognitive society — enabling communication, coordination, and delegation across diverse frameworks.