The Invisible Revolution: Why UX for AI Agents Might Be the Next Frontier
2026-02-28
Introduction: The Shifting Sands of User Experience
I recently attended a UX meetup featuring a talk titled:
"When bad UI isn't a failure: rethinking UX in games and complex systems."
A surprising amount of time was spent discussing the map system in Hogwarts Legacy. After the talk, I spoke to a friend who had played it.
He agreed with the speaker’s critique.
"It sucks."
And yet, the larger point of the talk was interesting: sometimes friction isn’t poor design, it’s intentional depth.
If you’ve read my earlier post UI/UX: How I Learned To Love Design, you’ll know I didn’t start out caring about design. I came from systems, structure, and logic. UX was something I learned to appreciate because I realised it wasn’t decoration, it was purpose.
At the end of the meetup, during Q&A, someone asked:
What impact will AI have on UX design?
The consensus was it wouldn't change the general approach as reliance on data and general adoption of inherent goal of friction reduction would continue to be simplified and basic for less friction as thats how designers are working.
But that question sparked an idea in my mind:
What happens when the user isn’t human at all?
If AI agents are designed to operate without human interaction, will we see websites and software with no human UX at all. Built purely for agents?
This isn't just hypothetical; it challenges the core of what we consider "good UX" and how we define value in digital products.
Part 1: Friction — From Enemy to Strategic Tool
Traditionally, UX has treated friction as the villain.
We:
- Reduce cognitive load
- Simplify decisions (Hick’s Law)
- Remove unnecessary clicks
- Streamline flows
And for most business software that is correct,but complexity is not always a failure.
Some tools are intentionally deep:
- Professional video editors
- Trading platforms
- Strategy games
They reward mastery which requires friction. Friction creates learning which creates identity which creates meaning. These systems reward mastery.
For humans, friction can be valuable and creates growth.
For AI agents?
Friction is failure.
An AI doesn’t care about micro-interactions, animation or experience delight.
It wants:
- Clean endpoints
- Structured responses
- Predictable schemas
- Machine-readable meaning
This forces a distinction:
Low Friction for Utility
Automation. APIs. Task completion.
Intentional Friction for Meaning
Learning curves. Depth. Emotional engagement.
The future of UX may depend on knowing the difference.
Part 2: When the Interface Is No Longer the Interface
If AI agents become users in digital ecosystems, something drastically changes.
The interface stops becoming visual and read on a screen. It becomes the structure behind it.
Structured Data has been treated as a technical concern, importand but secondary.
- SEO markup
- Database hygiene
- API design
In an agent focused world, that hierarchy flips the real "user experience" becomes:
- Your schema consistency
- Your API reliability
- Your authentication model
- Your machine-readable contracts
If an AI booking agent is choosing between two services, it won't care about which button looks nicer, or how much white space makes the text cleaner to read.
It will care whose data is clearer.
That changes advantage, shifting value from surface to structure.
This isn't a design shift it's a change of philosphy.
The Dual Future of UX
The future of UX isn’t a single, simplified disappearance into minimalism.
I think we will have a seismic split in designing for:
Humans
Rich, expressive, meaningful experiences. Sometimes simple. Sometimes deliberately complex.
AI Agents
Invisible, structured, machine-readable systems. Pure utility. Pure execution.
Designers and developers who understand both will have an edge, not because they can make prettier or more efficient interfaces but because they can design systems that:
- Appeal to humans clearly
- Architecture that machines can navigate
What fascinates me about this shift is it brings me back to where I started.
I began with systems, I learned to love design, now the two are converging.
It's recognising that products may soon serve two completely different users.
Man Vs Machine
Understanding both will define what comes next.