From A2A to MCP: How AI’s “Brains” are Connecting to “Arms and Legs”

As AI evolves, we can think of LLMs as the “brains” of AI systems, while agents and tools serve as their “arms and legs” – the components that interact with the world and execute actions. Between these lies a critical but often overlooked element: the protocols that act as a nervous system, transmitting signals, coordinating actions, and integrating feedback between AI models and their capabilities.

This talk explores how protocols like Model Context Protocol (MCP) and Google’s Agent-to-Agent (A2A) are enabling AI systems to extend beyond chat to practical action. We’ll explore how these protocols allow LLMs to access external tools, manage persistent memory, and collaborate with other agents.

We’ll examine where protocols excel in solving today’s challenges, and where they need to evolve to support increasingly complex AI applications. We’ll examine how the community can contribute to shaping this infrastructure layer, ensuring it supports the next wave of AI innovation.