The Approval Interface

There’s a new interaction pattern emerging in software, and I don’t think we’ve named it yet.

It started with Pinterest.

Back in the days when guided search launched, it felt like a small UX trick. You’d type one word, e.g. “living room”, and instead of a singular results page, you’d see a row of suggested refinements. “Cozy.” “Minimalist.” “Scandinavian.” Once you’d click one, new suggestions would appear instantly. You’d click again, and within a few steps, you’d arrived somewhere specific and beautiful without ever typing a second word. The whole UX was basically based on one single input. Everything else was just clicking “next.”

At the time, we called it guided search. What it really was, though, was something more fundamental: the interface had learned to meet you halfway. You didn’t need to know exactly what you wanted. You just needed to start.

I’m watching the same pattern evolve into something much more powerful at the moment.

At Lovable, I see this interaction pattern happen pretty much every day. A user types a single prompt, sometimes just a few words, and something gets built based on that. Once you see a first “result,” a suggestion appears that the AI has already anticipated based on your initial input. What I find interesting is that most users don’t ignore them but actually click them. The AI simply proposes the next action and all the user has to do is approve. And this loops all the time. Something else gets built → new suggestions pop up → they approve again.

If we zoom out, the product gets built almost entirely through a sequence of approvals.

And it’s not unique to Lovable. Look at how people use Cursor, ChatGPT, Notion AI, or any of the new wave of AI-native products. The interaction loop keeps compressing. The user’s job is less and less about constructing inputs and more and more about evaluating outputs. The interface runs slightly ahead of you, and you follow or you redirect.

Instead of initiating, we’re moving towards approving.

It’s a genuinely new interaction primitive, and I think it matters.