A2A (Agent-to-Agent protocol)

Definition

A protocol designed to allow AI agents built on different systems or platforms to communicate, exchange information and coordinate work.

Put simply

It gives different AI agents a common way to work with each other.

What boards and leaders should understand

Agents may increasingly operate across applications, vendors and organisational boundaries rather than inside one closed system.

That can make accountability more complicated. An outcome might involve your organisation’s agent, another organisation’s agent and several underlying systems, each with different permissions and responsibilities.

Questions to consider

  • If our agents interact with external agents, how will we establish identity, authority and accountability across those interactions?