A2A & MCP Workshop: Automating Business Processes with LLMs — Damien Murphy, Bench | AI Engineer

Damien Murphy presents a workshop on A2A and MCP, explaining how they can automate business processes using AI agents. He discusses the benefits of agent specialization, task delegation, and parallel processing with A2A, as well as the advantages of MCP as a standard interface for AI tools and context. He explains when to use and not use A2A and MCP, highlighting the importance of third-party tools and hidden complexity. The presentation includes a code walkthrough, demonstrating how to build a multi-agent system using A2A agents integrated with MCP, triggered by a webhook, and covers prompt caching and context management. The session concludes with an extensive Q&A, addressing questions about agent limitations, security, context management, LLM usage, and practical implementation challenges.

Outlines

Part 1: Introduction and Concepts

Part 2: Practical Application and Setup

Part 3: Agent Roles, Context, and Cost

Part 4: Code, Limitations, and Security

Part 5: Orchestration and LLM Selection

Part 6: Conclusion and Future Outlook

Sign in to continue reading, translating and more.

Open full episode in Podwise

A2A & MCP Workshop: Automating Business Processes with LLMs — Damien Murphy, Bench

AI Engineer

Part 1: Introduction and Concepts

Introduction to A2A, MCP, and Their Applications in Business Process Automation

Use Cases and Considerations for A2A and MCP

Benefits and Drawbacks of Using A2A and MCP

Part 2: Practical Application and Setup

Application Setup and Agent Functionality

Live Demo and Initial Q&A

Part 3: Agent Roles, Context, and Cost

Clarifying Agent Roles and Context Management

Context Size, Cost, and Benefits of Sub-Agents

Part 4: Code, Limitations, and Security

Code Walkthrough and A2A Limitations

Authentication, Authorization, and Security Considerations

Part 5: Orchestration and LLM Selection

Orchestration, Context Management, and LLM Selection

Managing Information Overload and Sub-Agent Communication

A2A vs MCP and Orchestration Control

User Context and Human-in-the-Loop

Context Management and Testing Strategies

Part 6: Conclusion and Future Outlook

A2A Complexity and Closing Remarks

A2A & MCP Workshop: Automating Business Processes with LLMs — Damien Murphy, Bench

AI Engineer

Part 1: Introduction and Concepts

00:15Introduction to A2A, MCP, and Their Applications in Business Process Automation

Introduction to A2A, MCP, and Their Applications in Business Process Automation

04:48Use Cases and Considerations for A2A and MCP

Use Cases and Considerations for A2A and MCP

09:11Benefits and Drawbacks of Using A2A and MCP

Benefits and Drawbacks of Using A2A and MCP

Part 2: Practical Application and Setup

15:08Application Setup and Agent Functionality

Application Setup and Agent Functionality

20:26Live Demo and Initial Q&A

Live Demo and Initial Q&A

Part 3: Agent Roles, Context, and Cost

27:51Clarifying Agent Roles and Context Management

Clarifying Agent Roles and Context Management

33:35Context Size, Cost, and Benefits of Sub-Agents

Context Size, Cost, and Benefits of Sub-Agents

Part 4: Code, Limitations, and Security

38:44Code Walkthrough and A2A Limitations

Code Walkthrough and A2A Limitations

45:31Authentication, Authorization, and Security Considerations

Authentication, Authorization, and Security Considerations

Part 5: Orchestration and LLM Selection

51:29Orchestration, Context Management, and LLM Selection

Orchestration, Context Management, and LLM Selection

57:33Managing Information Overload and Sub-Agent Communication

Managing Information Overload and Sub-Agent Communication

1:03:10A2A vs MCP and Orchestration Control

A2A vs MCP and Orchestration Control

1:09:12User Context and Human-in-the-Loop

User Context and Human-in-the-Loop

1:14:21Context Management and Testing Strategies

Context Management and Testing Strategies

Part 6: Conclusion and Future Outlook

1:19:26A2A Complexity and Closing Remarks

A2A Complexity and Closing Remarks