Is Structured Human Input the Missing Link in Agentic Work?

As autonomous agents become more prevalent, they frequently encounter tasks requiring human intervention. Currently, these pauses often rely on unstructured text, which creates friction and prevents seamless interaction between independent agent systems and clients. Ben Greenberg argues that the Agent2Agent (A2A) protocol provides a robust foundation for stateful tasks, but it lacks a standardized way for agents to request specific human input. By introducing structured input schemas, agents could define exactly what data they need—such as forms or boolean confirmations—allowing clients to render appropriate interfaces without custom adapters. This shift would move human participation from informal chat to a formal part of the workflow state. The proposal, currently under discussion in the A2A community, aims to create a more portable and interoperable ecosystem where human decisions are as legible to software as they are to the people making them.