sometimes i look at a really polished product and wonder if all that polish is actually helping.
more animations, more interactions, more personalized states, more ways to do the same thing.
at some point, “better UX” can start becoming more UX.
i’ve been thinking about the difference between making something feel delightful and just making it feel complicated.
where do you draw that line?
have you ever used a product and thought, “why the hell did they design it like this?” then later realized there was actually a pretty good reason for it?
could be anything, an annoying interaction, weird navigation, a feature that seemed unnecessary, whatever.
curious to hear the ones that changed your mind.
Been reading a couple of pieces about AI agents interacting with websites, and I’m not completely sold on calling agents “users.”
Functionally, I get the argument. An agent has a goal, encounters an interface, tries to complete something, and can fail because the interface is badly structured. That’s basically the definition of a user journey.
But UX is also about perception, comprehension, decision-making, emotion, trust, etc. An agent doesn't experience any of that. It parses information and takes actions.
Which makes me wonder if we're actually talking about two different layers of design.
The human still needs the interface to understand what’s happening and make decisions. Meanwhile, the agent increasingly wants structured information, clear labels, predictable states, semantic markup, APIs, etc. Interestingly, a lot of that overlaps with accessibility best practices anyway.
Another piece made this even more interesting by framing agents almost like a new distribution channel: your product doesn't just need to be attractive to humans, it needs to be legible enough for an AI to compare, evaluate and recommend it.
So where does this leave UX?
Are we heading towards a world where we have:
Human UX → designed for interaction and understanding
Agent UX → designed for interpretation and execution
And eventually the visual interface becomes almost irrelevant to the agent while becoming more important to the human?
Curious how people who actually work on product/design systems are thinking about this. Are you treating agents as another type of user already, or does that framing feel fundamentally wrong?
sometimes it feels like people associate great UX with big redesigns or flashy new features.
but i'm guessing a lot of the biggest wins come from really small changes that most users never even notice.
what's the smallest UX improvement you've seen that had a surprisingly big impact?
could be something as simple as changing button text, tweaking the navigation, reducing the number of steps, or anything else.
i'd love to hear real examples where a tiny change made a noticeable difference in usability or user behavior.
AI has gotten ridiculously good at generating UI screens, logos, illustrations, landing pages... basically anything visual.
but i'm curious, what's something AI still consistently gets wrong that stands out immediately to experienced designers?
not talking about weird hands or obvious image glitches.
i mean things that make you look at a design and think, "yep, AI made this."
would love to hear examples from people who use AI in their workflow regularly.