A2A in 50 Minutes: Building Interoperable AI Agents
summary
A2A is an open agent-to-agent protocol now at spec 1.0 (Java SDK CR1) that uses HTTP/JSON‑RPC, supports multimodal Agent Cards for capability discovery, and implements a full task lifecycle (submitted → working → completed/failed/rejected/canceled) with artifacts and webhook callbacks. Demo code uses Quarkus, LangChain4j, an Ollama/Gemma4 local LLM and mixed Java/Python servers that interoperate, but the stack is bleeding edge (SDK mismatches, Red Hat TCK revealed Python gaps) so it works for experiments, not immediate production.
AI-written summary · about
chapters (9)
- 0:00Introduction and speaker
- 3:36What is A2A protocol
- 9:14Agent cards and demos
- 12:06Task lifecycle explained
- 21:57A2A vs MCP comparison
- 25:27Code walkthrough and demo
- 37:46A2A use cases and ecosystem
- 40:00Current limitations and future
- 42:28Q&A and resources
related talks
39mA2A in 45 Minutes: Building Interoperable AI Agents
27mA2A in 25 Minutes: Building AI agents that communicate
31mA2A in 30 Minutes: Building Communicating AI Agents
34mWhat can A2A do for you and what can you do for it
10mKeynote: The Agent2Agent (A2A) Protocol
16mAgent2Agent Protocol and the Future of Agent Collaboration
02h 59mStandards and AI Agents: an Overview of MCP, A2A, ADK and More
01h 02mAI in Java: We've Built a Server-Side Enterprise Google A2A for Jakarta EE Devs.