Lovable, Bolt, Base44: I've been building apps for 15 years, and these tools are good news for every app builder

AI app generators don't make no-code app builders obsolete. They speed up the first version. Publishing to the stores, running an app day to day, and keeping it alive: that's a different job.

Since 2011, I’ve worked at GoodBarber, a Mobile App Builder. Our job: to put a tool in people’s hands so anyone can have a mobile app of their own, then help them build it, publish it, and keep it running. I’ve watched the waves go by, the breakthrough announcements, the “this changes everything” claims that changed very little. Over time, you learn to tell a real shift from noise.

Lovable, Bolt, and Base44 are a real shift.

Bolt.new says it went from zero to $40 million in annual recurring revenue five months after launch. Base44, built solo and with no outside funding, was acquired by Wix about six months in. The Collins Dictionary named “vibe coding” its 2025 Word of the Year. This is not a gadget.

But the numbers aren’t what matter most. What matters is that one evening, someone thinks “instead of watching a series, I’ll build something for my business,” and this time, they actually get off the starting line. The blank page, that invisible barrier that used to stop people before they even began, is receding further. People who would never have taken the step are taking it.

That’s exactly what we’ve been trying to make possible for fifteen years. Hard, under those conditions, to see it as a threat.

Are app builders dead?

No. AI app generators and no-code app builders overlap, but they don’t work the same moment in an app’s life.

The former have shortened the path from an idea to a first working version. In June 2026, Lovable said close to a million new projects were being created on its platform every week. That’s not a million apps ready to live in the world. It’s a million beginnings.

And a beginning, however exhilarating, quickly raises a simple question: now what? How do I get this thing to actually exist in my users’ hands, and still work six months from now?

Faster creation doesn’t remove the difficulty. It moves it one step further along. Generating a prototype and making a product last are two different jobs.

My colleague Muriel sat with this exact question for a year, from the marketing seat: will AI kill no-code app builders?

Will AI kill no-code app builders? What the industry won't tell you

Her answer is more interesting than a yes or a no.

Four problems the first version doesn’t solve

I’m not going to give you a list of what Lovable, Bolt, or Base44 can and can’t do: it changes too fast for any snapshot to stay accurate. I’d rather tell you about the road we travel every day, the one that runs from a demo that works to an app that’s published, operated, and maintained. It’s long. Here is where it goes.

1. Distribution

Publishing to the App Store and Google Play isn’t a button you press. There are developer accounts, certificates, the formats each store expects, privacy declarations, and the back-and-forth that sometimes follows a rejection.

Concretely: over the past twelve months, Apple rejected roughly 42% of the first-time iOS submissions handled by our team on the first pass. After fixes and resubmission, 91% of those rejected builds were eventually approved.

Source: GoodBarber publishing team internal CRM tracking, trailing twelve months, figures as of April 2026.

The goal isn’t to promise an app will never be rejected: nobody escapes that entirely. The goal is to prevent the known reasons, understand the ones that happen anyway, and know how to carry the case through to the end.

2. Mobile integration

A mobile app isn’t a website that shrank to fit a smaller screen. It talks to the operating system, and it has to respect the stores’ rules.

A subscription to digital content, for instance, has to go through StoreKit or Google Play Billing. A notification has to arrive at the right moment. Scrolling has to stay smooth, the haptic buzz has to land just right, permissions have to be asked for cleanly. Insider stuff, sure. But the user feels it instantly: it’s what puts an app among the ten they open every day, or among the ones they delete the next morning.

A first version shows that a feature exists. It doesn’t yet tell you how that feature will behave across devices, inside the stores, and over time.

3. Day-to-day operation

Once it’s published, an app starts to live. Someone has to update the content, manage users, send notifications, track orders, change a price, look at what’s working.

So the question is no longer just “can I build this feature?” It becomes: “who’s going to run it day to day, without rebuilding the product every time something changes?”

That’s where a solid back office earns its value. Not on demo day. On day 100. Then day 500.

4. Maintenance and compliance

iOS and Android move. Store rules change. Dependencies need updating. And privacy, consent, and your users’ data keep applying long after launch.

Maintenance is what keeps an app installed last year still working today. There’s nothing spectacular about it. It’s exactly what decides whether a first version becomes a product that lasts.

Distribution, integration, operation, maintenance: these four are set up at creation, but they only reveal themselves after the first generation. AI compressed that first step. It didn’t make the ones that follow disappear.

AI app generator or no-code app builder: which tool should you choose?

There’s no single answer, because these tools don’t serve the same projects.

Want to explore a custom idea quickly, spin up an internal tool, test an interface? Lovable, Bolt, and Base44 are remarkably efficient entry points. Their strength is the speed with which an intention becomes something you can see, touch, and change.

Is your project a mobile content or commerce app, meant to be published on iOS and Android and then run every day by a non-technical team? That’s where a Mobile App Builder becomes more relevant. At GoodBarber, everything sits in one place: the question is no longer how to generate the app, but how to cut the number of pieces you’ll have to assemble and maintain afterward.

Is your project a game, a complex marketplace, a product built on very specific business logic? Custom architecture and development is often still the most sensible choice.

So the right tool isn’t the one that puts the first screen on the glass fastest. It’s the one whose strengths and limits match what you’ll still have to do after that first screen.

Start, then look past the beginning

That’s why I don’t see AI app generators as a threat, but as good news. They bring people to creation who would never have come. They make ideas easier to try, and first versions easier to share. And along the way, they push all of us to raise our standards.

So if, one evening, you decide to build something instead of watching a series: start. Really.

Just keep a second question in your back pocket, to ask right after the first: if this idea works, what will it take to publish it, run it, and keep it alive?

That’s where the interesting part really begins. And that road is learned, and it’s shared: nobody has to walk it alone.