Something extraordinary happened last week that few people recognised for what it truly was.

In mere months, two protocols have emerged, each with an unassuming name but profound consequence. Anthropic’s Model Context Protocol (MCP) arrived first, quietly announced late in 2024. Soon after, in the brisk spring of April 2025, Google's Agent2Agent (A2A) appeared, surrounded by eager partners such as ServiceNow, Salesforce, and SAP. They arrived quickly—not through lengthy committees or meticulous bureaucracy, but through collective recognition of necessity.

This isn't merely a better way to code. It's a fundamental rewiring of what's possible when intelligence doesn't just flow through systems but between them—creating new forms of value we've only begun to imagine.

MCP is a universal connector, a simple yet elegant means by which AI agents may communicate with enterprise systems and data.

A2A complements it by allowing these agents, distinct yet sympathetic intelligences, to converse seamlessly with one another.

Together, they suggest not merely improvement, but transformation—an awakening of collaborative intelligence reminiscent of life's own inclination towards co-operation and specialisation.

And one more thing: The organisations that recognise this shift and act on it now will create insurmountable advantages over those that don't.

If this moment reminds me of anything, it's that day in 2007 when Steve Jobs pulled the iPhone from his pocket. Many technical experts dismissed it:

"It's just three existing technologies combined."

They completely missed the point. When transformation happens, it rarely looks transformational at first glance.

The Protocol Revolution — Explained Simply

Let's be honest: technical protocols are about as exciting as watching paint dry—unless you understand the tectonic shift they represent.

Think of A2A and MCP not as specifications, but as a new social contract for software. Together, they form a new foundation for enterprise AI:

MCP is essentially USB-C for intelligence

It standardises how AI systems connect to tools, data sources, and knowledge. Before MCP, connecting a language model to your company's databases, APIs, or tools required custom engineering for each connection—an M×N integration problem where complexity exploded with each new system.

MCP simplifies this to an M+N problem: build connectors once, use them anywhere.

A2A is like inventing meetings for software

It creates a standardised way for independent AI agents to discover each other's capabilities, coordinate on tasks, exchange information, and track progress. Instead of hardcoding which agent does what, agents can dynamically find and collaborate with each other based on the task at hand.

If you're struggling to grasp the significance, consider this metaphor: