Users don’t care about your product.
Or how long it took to build.
They care about one thing:
“Will this work for me?”
If your product doesn’t create certainty fast enough,
they leave.
And choose a competitor that does.
Knowing why something works… not just what looks good.
AI can generate.
But it doesn’t know when it’s wrong. It doesn’t know when to stop… or when to go manual.
Designers do.
The best work right now?
AI speed + human judgment.
It’s not AI vs us.
It’s AI + us.
You can try teaching AI what “good design” looks like…
prompt after prompt… hoping it gets there.
Or you let a designer handle it.
Because design isn’t just output.
It’s context. Trade-offs.
(Continued...👇)
@BenjaminUIX For the non-designers who have to rely on designers it gives amazing results.
But I'm working in a company where all of my work shifts towards giving design "Human touch" because in the end people realize without "human" the design look just meh.
@BenjaminUIX You can provide assets, better context about each section in prompts... & It will generate better results than what you have here.
But to polish the design further, you can't rely on Claude.
It's a paradox... just one more prompt & it'll be done but it never does.
Claude sucks at iteration...
The second you try to iterate,
you lose time instead of saving it...
It falls apart & you reach nowhere with the limits it imposes.
Is it just me… or does it make simple edits harder than they should be?🤷🏻♂️
@AnkitKu2175@yahyavision This is funny on so many levels 🤣🤣
Finally companies are realizing that it's not worth wasting credits to do tweaks to get polished results.
AI provides a solid start, but i haven't seen any pixel perfect results by AI.
@kylegawley Founders are now realizing this problem, & most of them are learning form their own mistakes.
Use AI as a helper, speed up the workflow but not so much that you can't review anything.
@kylegawley Absolutely, Unless you are 60-70% sure of what you actually want, you end up losing money.
Better prompt requires certainty on human sides... Only then you get the most out of these tools.
Claude is just a modern gambling machine...
It doesn’t just help you work,
It hooks you.
It gives perfect outputs, sometimes
But mostly “almost there.”
You end up tweaking, endlessly.
You’re not building.
You’re pulling a slot machine lever, hoping this prompt finally hits.
@ucfmkr@jerrod_lew That's exactly my point, the effort has shifted.
It's no more about execution, it's about specifying the requirements.
Yes, AI can generate results in a few mins, but the effort you need to put to specify every detail... It's still there.
@LukeNowakPL@tranmautritam Thisssss... 👏
I use Claude in my workflow, everyday. It's a good assistant.
But understanding how humans interact with products to get conversion has always been an iterative process. AI can definitely speed up a lot of things, for sure, but can never replace us.
@Abmankendrick Also, If images used below won't be too light you can go with B. But what if there's an image that is white, would it affect the readability too much, even with the overlay??
That's why we don't see aesthetically pleasing UIs in products that have been in the industry for years.
@Abmankendrick Tbh, it's just about personal preference.
Readability is not an issue in the B as you can make the gradient a bit darker below the text to increase the contrast for the text.
Aesthetically, B looks better. But it is not necessarily better if you know what i mean.