A few times now, a founder has opened a call by sharing their screen and showing me something they built in Lovable or v0. I can usually tell within about five seconds that it wasn't designed. Not because the tool is bad. Because the colour choices and the font pairing have a certain look, and once you've seen it a few times, you recognise it instantly, the same way you can spot stock photography before you've read a word of the copy.

The tool was never the bottleneck

Here's the thing that took me a while to say plainly: the problem isn't the AI. A vibe-coding tool is only as good as the prompts and decisions steering it, and most founders aren't positioned to make good design calls under pressure, in a tool they opened an hour ago. That's true whether the output lands in Framer or in Lovable. Neither tool fixes a taste problem. If you're not the right person to be making those calls, the answer isn't a better tool, it's someone who is.

And that someone doesn't have to be a Framer designer specifically. A good designer can vibe-code something decent. But in practice, building in Framer is still the path to the highest-quality output for the least ongoing pain, and here's the actual reason why.

What changing one thing later actually costs

In Framer, the site is structured, CMS-backed, built in blocks. Swap a paragraph, change an image, update a price, and nothing else moves. That's the entire value of a CMS: the change stays where you made it.

A vibe-coded site doesn't have that structural floor. Change one thing and you have to check that the change didn't ripple somewhere you didn't intend, that responsiveness still holds, that nothing quietly broke two sections down. A lot of real work goes into a vibe-coded site staying correct after launch, even though the pitch is that it's effortless. The effort doesn't disappear. It just moves from the build to every edit after.

Where vibe coding is honestly the right call

Fast, and cheap upfront, and I won't pretend that doesn't matter. If a founder has no runway, the honest advice isn't "build it properly in Framer anyway." Even a Framer plan is a recurring cost, and that's a real number for someone counting months.

My actual advice in that situation, and it goes against my own business: it's better to have no site than to ship one that visibly reads as vibe-coded. An obviously AI-generated site in front of an investor or a first customer can cost you more credibility than having no site at all. It signals that you either didn't notice or didn't care, and neither is the impression an early-stage company can afford.

When there's runway, build the real one

If a founder has enough runway to do it properly, that's when a real Framer build earns its cost. Something you can put in front of an investor, or a customer, and have it read as a legitimate, high-quality product, not a fast experiment. That's the site that does work for you in a room you're not in.

So the actual decision isn't Framer versus vibe coding as a feature comparison. It's: do you have the runway to do this properly, and can you get someone with design judgment on it, whichever tool they use. If the answer to either is no, wait, or keep it minimal, rather than shipping something that undersells you the moment someone looks closely.