Main Question: What are some of the ways to structure the communication between two agents? - What exactly did openai change with the Responses API? - A2A protocol? - MCP? (previous issue?) - What are the limitations of current approaches? - From first principles, what is communicated between two agents? - How will agent to agent be different from agent to user? - Agent Frameworks? AgentJo and Langchain Agent, n8n -> how do these frameworks solve the problem?