a few weeks ago i was reviewing a product hunt launch.
the product itself was actually interesting. solid idea, real problem,
working demo. but i closed the tab in about eight seconds.
not because of the product. because of the landing page.
something about it just felt off. the spacing was slightly wrong.
the copy had that particular rhythm that comes out of chatgpt prompts.
the components looked like they were assembled rather than designed.
i didn't consciously think any of this. i just felt it and left.
that's the vibe coding problem.
what is vibe coding exactly
vibe coding is building software by prompting ai tools like cursor,
copilot, or claude to generate your frontend for you.
it's fast. it's genuinely useful. and it leaves patterns.
ai generated frontends tend to have specific tells. inconsistent
spacing that's almost right but not quite. button hierarchies that
don't follow any real visual logic. copy that uses certain sentence
structures over and over. color usage that looks picked from a default
palette rather than a considered system. components that work
individually but don't feel like they belong together.
none of these things break the product. but they signal something
to the people looking at your site. especially developers and
technical users who have trained eyes from seeing hundreds of
product hunt launches every month.
the signal they pick up is: this was assembled quickly by ai
and nobody reviewed it with fresh eyes.
why these patterns are detectable
the patterns are consistent enough to flag programmatically.
spacing inconsistencies show up in computed styles. generic class
naming conventions appear in the source. certain component structures
repeat in ways that human designers don't produce. copy patterns
emerge that correlate with common prompt outputs.
none of these signals alone means anything. but combined and scored,
they paint a picture.
what humanon does
humanon takes your site url, scans the live frontend and code signals,
and returns a vibe score with a breakdown of exactly which patterns
it detected and why they matter.
not a general "this looks ai generated" verdict.
specific flags. specific reasons. specific one-line fixes for each one.
the goal is simple. you run humanon before you launch.
not after the first comment on product hunt tells you your
landing page looks like every other cursor app.
i built this because i was the person getting that comment.
if you want to try it, link is in the comments.
let me know what score your site gets.
Top comments (3)
Interesting concept, but I wonder how much of “vibe coding” is just “bad design patterns that already existed before AI.” Feels like the harder part will be separating AI signals from just inexperienced frontend work.
tried this on my own site and was genuinely surprised by what it flagged. all the small effects , color gradients, etc.. that I thought normal was not.... and this was the culprit pushing my users away from the site
Some comments may only be visible to logged-in visitors. Sign in to view all comments.