When everything is easy to build, taste becomes the bottleneck
I’ve been spending more time vibe coding recently, and I’ve started to question something I initially took for granted. Most of the conversation around vibe coding is about speed. Like how quickly you can go from idea to prototype, or how fast you can iterate. And to be fair, that part is real. The barrier to building has clearly dropped.
But the more I use these tools, the more it feels like speed isn’t the limiting factor anymore.
The real constraint seems to be taste.
what do you choose to build?
what do you keep vs discard?
what actually feels right vs just “working”?
what is genuinely useful vs just impressive in a demo?
I’ve had plenty of cases where I built something quickly that technically worked,
but still felt off. And other cases where I spent more time refining something simple
until it felt obvious and clean. The thing with Vibe coding is that it makes output abundant.
But that also means it’s easier to produce a lot of mediocre output very quickly. So I’m starting to think the core skill isn’t prompting or tooling, but judgment.
Almost like product taste, it’s about knowing what should exist, what shouldn’t, and when something is actually “done.”
Curious how others here see it:
How has vibe coding changed the way you decide what’s “good” vs just “done”?
Replies
For commoditized apps, I hold high expectations like AI dictation tools, where the bar is already set.
But for novel apps breaking new ground in their category, I'm perfectly fine if they simply get the job done.
taste was always the bottleneck tbh, vibe coding just made it impossible to ignore. before you could blame slow progress on implementation difficulty but now theres nowhere to hide
I completely agree with you!
Vibe coding has certainly made a lot of things easier, but it doesn't mean that everyone is capable of building something cool.
If you're only interested in what "works," then it's fine.
But if you really want to differentiate from your competitors and increase your chances of success, taste still matters.
Of course, I'm not 100% neutral 'cause I build brand identities and websites with my partner lol
But I like to believe that our work still matters and can make a real difference (hopefully).
Hello Aria
This hits at something I think about constantly. When frameworks, no-code tools, and AI copilots commoditize execution, the differentiator becomes the quality of the question you're asking, not how fast you can build the answer.
Building Hello Aria, our AI assistant in WhatsApp and iOS (launching PH April 10th), we've noticed this firsthand. Anyone can ship a "personalized AI assistant" today in a weekend. What's hard is tasting the difference between an interaction that feels helpful vs. one that's technically correct but contextually off. Taste is just accumulated intuition about what matters to real humans.
The builders who develop real taste for their users are going to create moats that vibe-coding can't shortcut.
It's taste and UX. How to optimize the onboarding flow for new users? How to declutter the interface? And as someone has said here: taste has always been the bottleneck :D
AI is great for fast prototyping, but to stick out you have to be unique - and that is not easily possible if your tool is mostly based on mathematic possibilites.
Somewhat related -- products are fast and easy to ship, so what truly differentiates a good first product from a bad one?
So many people say nowadays that MVP is dead and I do have to agree. more about MLP and MVE, which puts at the core the product being likable by the mass from the very first launch. So, much deeper architecting of the UI/UX from the get-go.
If any non-coding person can build any functioning product, the industry gets spammed with so much noise (perhaps product hunt is also going through the same problem nowadays). Becoming a signal requires something more than good functioning features.
The other I think is a real technological/data moat, but that comes after MLP and data accumulation...
I agree, but I would frame it differently. I would say the bottleneck is signal. What do people want? What messages resonate? And what's going to make a difference to user behavior?
I completely agree with this. When everything becomes easier to build, the hard part shifts from “can I make it?” to “should this exist, should this be here, and does it actually help the user?”
I’m building Traction, and this has come up a lot. It’s very easy to keep adding pages, modules, buttons, dashboards, AI sections, and shiny features because the tools make it possible. But possible does not always mean useful.
Taste becomes deciding what to remove, what to simplify, what should be one workflow instead of five pages, and what actually helps the user get the outcome they came for.
For Traction, that outcome is not “more features.” It’s helping a business turn visibility, content, leads, and follow-up into booked revenue. I think that is where taste really matters: not just making the product impressive, but making the path through it obvious.