For a long time, I evaluated UX success by how smoothly users moved through a flow.
Did they click the right thing?
Did they complete the steps?
Did they reach the end without friction?
That mindset served me well for years.
But once I started working on AI-driven products, I realized something subtle had changed:
Users weren’t trying to complete flows anymore — they were trying to express intent.
And clicks turned out to be a very weak signal of that.
Clicks Are a Proxy, Not the Goal
Traditional UX treats clicks as evidence of progress.
If users click:
- the button is clear
- the flow works
- the design is successful
But clicks only tell us what users did — not why they did it.
In AI products, the “why” matters far more than the interaction itself.
A user clicking through five screens doesn’t mean the product helped them. It just means they complied with the interface.
AI changes the game because the system is expected to interpret, not just execute.
Intent Is Messy — and That’s the Point
One thing I had to accept early on:
User intent is rarely clean or complete.
People don’t arrive with perfectly formed requests.
They arrive with:
- partial goals
- evolving context
- uncertainty about what they actually need
Traditional UX tries to clean this up early:
- force a choice
- lock a path
- reduce ambiguity
AI-native products work better when they do the opposite.
They allow intent to:
- emerge
- change
- be refined mid-interaction
Designing for intent means designing for conversation, not completion.
When Flows Get in the Way
I’ve seen well-designed flows become obstacles in AI systems.
Not because they were confusing — but because they demanded decisions too early.
Questions like:
- “Choose one option to proceed”
- “Select the correct category”
- “Confirm before continuing”
These make sense in deterministic systems.
In AI systems, they often interrupt thinking.
The user hasn’t decided yet.
They’re still figuring it out — sometimes with the help of the system.
Designing for intent means letting the system do some of that work, instead of pushing it back onto the user.
The Role of the Interface Quietly Changes
Once intent becomes central, the interface takes on a different role.
It’s no longer there to:
- guide users through steps
- enforce order
- prevent deviation
It’s there to:
- capture intent
- reflect understanding
- allow correction
Good intent-driven interfaces feel less like forms and more like mirrors.
They show users:
“Here’s what I think you mean — tell me if this is wrong.”
That single shift changes how trust is built.
Measuring Success Gets Harder (and More Honest)
One uncomfortable side effect of designing for intent:
success becomes harder to measure.
Clicks, completion rates, and funnels stop telling the full story.
Instead, you start caring about:
- how often users refine rather than restart
- how quickly they reach clarity
- how much steering the system requires
These aren’t always clean metrics.
But they’re closer to real value.
And they force more honest product conversations.
What I’m Still Learning
I still catch myself defaulting to click-based thinking.
I still ask:
- “What’s the primary action here?”
- “What’s the next screen?”
Sometimes that’s useful. Often, it’s a habit.
Designing for intent requires slowing down and asking a different question:
What is the user actually trying to accomplish — even if they can’t say it clearly yet?
I don’t always get it right.
But when the product stops asking users to “click correctly” and starts helping them think, the experience changes noticeably.
Final Thought
Clicks are easy to design for.
Intent isn’t.
But if AI products are meant to reduce cognitive effort rather than add to it,
intent has to become the primary design material.
Everything else is just an interface detail.