David Rant

As AI agents become a core part of our digital fabric, organisations are seeking to build the infrastructure needed for a scalable, secure, and composable AI future. Unlike traditional tools, agents can plan, reason, and use tools autonomously, with or without human intervention. However, enabling these autonomous entities to collaborate reliably and responsibly across diverse systems and with each other requires robust communication frameworks.

This talk serves as a developer's guide to understanding and implementing two foundational protocols for this new era: the Model Context Protocol (MCP) and the Agent2Agent Protocol (A2A). We will demystify these essential building blocks that support the agentic enterprise.

We will explore how MCP enables AI agents to securely and efficiently interact with enterprise systems and data sources, providing the necessary context for agents to reason with live data and execute actions. By defining tools based on Remote Procedure Calls (RPC), MCP allows the underlying Large Language Model (LLM) to access API specifications and reason with them without needing custom logic for each integration.

Following this, we will delve into A2A, demonstrating how it facilitates seamless collaboration by allowing AI agents to communicate, securely exchange information, and coordinate actions with each other.

Together, MCP and A2A are laying the foundation for scalable, multi-agent ecosystems and enabling organisations to move from siloed agents to orchestrated workflows.

The focus will be on enabling usability, enhancing flexibility across diverse systems, and ensuring robust security and governance for agent interactions.