Agentic UX
AI & Emerging Tech
Interfaces for AI that takes actions on your behalf, not just answers questions.
In practice
Agentic UX is the design of experiences where an AI agent plans and carries out multi-step tasks — booking, researching, building, buying — with the user supervising rather than operating. The hard problems are trust and control: how much the agent explains, when it asks, how you stop it, and how you know what it did.
Questions
What is Agentic UX?
Agentic UX is the interface layer for AI that acts — planning steps, using tools and finishing tasks — where the person delegates and reviews instead of doing each step.
Why do you use Agentic UX?
Delegation is the point of an agent. If the interface makes you babysit every step you have lost the benefit, and if it hides everything you cannot trust it.
How do you use/apply Agentic UX?
Show the plan before the action, make interruption instant, surface what changed, and give the person a clear undo. Ask permission for anything irreversible or expensive.
When do you use Agentic UX?
For tasks that are long, repetitive or spread across tools — and only once the agent is reliable enough that supervising beats doing.
Who uses Agentic UX?
Product and interaction designers working on assistants, automation and any product where AI acts in the world.
