Why You’re Addicted to Lovable Before You Even Hit the Paywall

The Covid pandemic helped turn the company Zoom into a verb. It was the standout success of that time, accelerated by the tailwind of everyone being stuck at home.

There's a chance that Lovable reaches similar category-defining heights with their take on "vibe coding"—essentially helping non-developers build products by prompting AI models.

But when I recorded exactly what I did during a realistic 100-minute build, nearly 60% of my messages were simply pointing out bugs. That sounds like product hell.

And more importantly, how have they used this to upsell their premium service so effectively, that they've grown to $100m ARR in just 8 months?

Welcome to the product design casino.

As you use vibe-coding apps like Lovable, one thing becomes clear: you'll spend the vast majority of your time doing nothing.

You're not incentivised to think about problems. Instead, you just ask the AI to "do it again, but better", and wait. Maybe this is what being a Roman emperor felt like.

Consider the following:

This is just an observation of my own behaviour, but because of those three factors, I quickly settled into my role as a lazy dictator and did nothing.

And although I was often sat for 5+ minutes, doing nothing, I wasn't thinking—there wasn't a strong enough incentive to do that.

Even worse, you might start multi-tasking, by watching Netflix or aimlessly scrolling Twitter. That kind of 🗯 Context Switching can be really damaging to both creativity and productivity, even if it feels efficient at the time.

What worries me is that this periodic downtime, combined with delegating key product decisions to AI, may gravitate designs towards mediocrity.

So, what can these vibe coding apps do to incubate creativity while their AI is working? Surely there's a better and more productive way to use the right-hand side of the screen?

• Could it be used to write notes that get turned into tasks later?
• Could you continue previewing the app and log bugs into a "to do" list?
• Could it show you more simple "this or that" design decisions, for you to quickly select?

But maybe it's designed perfectly for what it is...

Yes, Lovable has many characteristics of a casino game—but most of those are just byproducts of the fact that they use AI. 

So what's the takeaway here? Be a casino? (No).

Instead, think about my 100 minutes as a new Lovable user.

What Lovable have done so effectively with their free plan, is:

This may sound obvious, but it would have been so easy to just encourage users to plan more, to use their credits sparingly, and to start building something really ambitious.

In fact, if you designed Lovable with the assumption that everyone was building a scalable product, this is probably what you'd do.

Instead, they're controlling what happens in those first 5 minutes.

And that's the takeaway here: those first few minutes don't actually need to represent the stresses of the full experience—you want to showcase the best thing it can do.

Lovable aren't building their own AI models, they're building the experience layer around them.

As the underlying AI improves, some of the current issues will fade away—particularly around response times and error rates.

But right now, what is Lovable?

Whilst I'm sure that's technically feasible, the actual experience right now has characteristics and psychology that lean more towards prototyping—and it works really well for that.

But consider the dilemma of what you actually want from a rapid prototyping tool, compared to one that builds scalable products:

It's not that Lovable is designed poorly—it's actually excellent at what it does. I'd argue that it's best-in-class.

The point is that it's so addictive, enjoyable, and successful because it makes building ideas simple. Every aspect of the experience has been fine-tuned to make you think less.

But has great UX ever been created by thinking less?