Designing outcome-first “Do it for me” products
Agentic UX is the shift from software and systems that help you do tasks to software that does tasks for you. Instead of giving users a long series of screens and options, agentic experiences let people state an intent (e.g. “book the cheapest train that arrives before 9am” or “draft a reply that matches my tone”) and then the system plans and executes steps across tools, forms and services.
That creates a new UX problem: users aren’t just navigating, they’re delegating responsibility to a third party and only confirming its recommendations. Good agentic UX makes delegation feel safe by clearly showing what the system is doing, what it’s about to do, what it needs from the user, and how to undo or correct outcomes. It also changes settings from static preferences into lightweight controls that guide behaviour (e.g. “always ask before purchases” or “select a preferred supplier”).
When we’re developing agentic UX, we are always looking to take the mystery out of the experience for the user by making the plan visible, requiring confirmation for irreversible steps and always providing an easy undo.
