Skip to main content

What is an agentic UI?

An agentic user interface lets software act on the user's behalf: it plans steps, calls tools and reports back. The user no longer clicks every step, so the interface must show what the agent does, why, and how to stop it.

Three differences from a classic UI​

  • The system takes initiative. A classic UI waits for a click. An agent may start work on its own.
  • Results are uncertain. The same request can give different answers, so the UI must show confidence and sources.
  • Actions have side effects. An agent can send, change or delete things, so the user needs control and a way back.

Learning goals​

After this module you can:

  1. name the three differences above;
  2. choose a pattern for dry run, approval, progress and recovery;
  3. judge an agentic UI on trust and control.