Twice as many new apps, almost the same number of downloads. Why getting noticed is now the hard part of building with AI, and what I built to help.
Last week I wrote that when anyone can build an app in a weekend, most new apps will find no one. I said it was a guess.
It isn’t really a guess any more.
About 600,000 new apps arrived on Apple’s App Store in 2025, up 30% on the year before. In the first six months of 2026, another 560,000 arrived. That’s nearly a whole year’s worth in half a year. Those numbers are Sensor Tower’s estimates, reported by The New York Times in July.
Here’s the part that matters if you’ve built one of them. Over the same six months, App Store downloads grew 2%.
Twice as many apps. Almost the same number of people downloading them.
And that’s only Apple’s store. Plenty of apps built with Lovable, Bolt or Replit never go near an app store. They’re websites with a login, shared on LinkedIn or in a Facebook group. Reported figures for Lovable alone run from 100,000 to 200,000 new projects started every day. Most of those will never be finished, let alone launched. But the ones that are launched are all asking the same people for the same few minutes of attention.
Getting noticed is half of it. The other half is getting someone to actually sign up.
Think about what you’re asking a stranger to do. Sign up to an app they’ve never heard of, built by someone they don’t know, probably in a few weeks, and type in their email, their business details, sometimes their customers’ details.
They’re right to hesitate. In a 2025 study, Veracode tested AI-generated code on security-sensitive tasks across more than 100 AI models, and 45% of the results contained a well-known kind of security flaw. The same year, a researcher scanned 1,645 apps built with Lovable and found about 1 in 10 let anyone read or change data they shouldn’t have, including usernames, emails and payment status. (Lovable disputes treating that as a flaw in Lovable itself, and says each app’s owner is responsible for protecting its data.)
None of that means your app is unsafe. It means a stranger can’t tell either way. And when they can’t tell, most of them close the tab.
So I built two things into Verilay, and they work together.
A directory of apps built with AI tools: Checked apps. Every app in it has been through a Verilay scan. That’s the only way in. Visitors can search by what they need (“bookings”, “quotes”, “home build”), filter by the tool it was built with, open the app, or contact the person who built it.
A stamp for apps that pass. If the scan finds no critical issues and grades your app A or B, your listing carries a green stamp, Checked · No critical issues, with the date. You also get the same stamp to put on your own site, with copy-and-paste code for your footer or your GitHub README.
A few details I cared about:
I want to be straight about this, because a badge that overpromises is worse than no badge.
My own apps are the first in the directory, and it’s still small. That’s fine. I’d rather have ten real, checked apps than a thousand nobody looked at.
If you’ve built something with Lovable, Bolt, Replit, Cursor or any other AI tool, run a free scan and list it. And if you think I’ve got the stamp wrong, too strict, not strict enough, or missing something you’d want to see, tell me at [email protected].