Will AI kill no-code app builders? What the industry won't tell you
If AI writes apps now, are app builders dead? Honest answer from someone who’s watched this industry through several supposed apocalypses — and who thinks this one is different, just not in the way you’d expect.
Every week, someone asks me if AI just killed the tool I work on. Sometimes it’s a founder in my LinkedIn messages, sometimes a friend at dinner who’s just watched a demo on X and wants to know, gently, if I’m worried about my job.
I started working in tech in December 2014, which has given me time to attend a few of these funerals. In my first years on the job, the recurring story was that nobody downloaded apps anymore — every few months a new study came out about how the average person installed zero new apps per month, and every few months someone concluded the category was finished. Then chatbots, around 2016 — every conference had its slide about how nobody would ever tap a screen again, we’d all just talk to bots. Then progressive web apps were going to replace native apps for good. I watched that one from the front row: native apps didn’t die, and PWAs didn’t fail either — they just became two ways of shipping the same app. Today we build both, and nobody remembers there was ever supposed to be a war. That’s the pattern, every single time: the demo is spectacular, the category is declared dead, and two years later the category is still there, slightly transformed, having absorbed the thing that was supposed to replace it.
So my reflex, when the AI question started coming, was to file it with the others. That reflex was wrong, and I want to be precise about why — because the answer I’ve landed on after a year of sitting with this is more interesting than either “we’re doomed” or “nothing to see here.”
This time the demo is actually good
Here’s what makes this wave different from the chatbot one: the demos deliver. You type a sentence, and something real appears — screens, navigation, working buttons, a colour scheme that doesn’t hurt. I know because I’ve done it. Earlier this year I spent my evenings building with an AI coding tool, me, a marketer who had never written a line of code, and the first hour genuinely felt like a small miracle. I’ve written about that experience separately; the short version is that the magic is not fake.
That’s why I don’t roll my eyes at the question anymore. When something can produce in thirty seconds what used to take a team three weeks, “is this the end of your category” is a reasonable thing to ask. The people asking aren’t naive. They’re extrapolating from what they saw.
The problem is what they saw. A demo shows you an app being born. It cannot show you an app being alive. And after years of working next to the people who deal with the second part, I can tell you the two have almost nothing in common.
What I know because I’ve watched it, not because I’ve read it
The marketing seat at an app builder is a strange observation post. You don’t write the code, but you see everything that goes wrong around it, year after year, because it lands in the support tickets, in the release notes you have to announce, in the customer emails.
You see what an App Store rejection actually does to a small business owner three days before their launch event. You see what happens every autumn when a new iOS version ships and a feature that worked for four years suddenly doesn’t. You learn that push notifications are a certification, payments are a compliance project, GDPR is a lawyer’s afternoon at minimum, and that an app which was perfect at launch quietly rots if nobody recompiles it, resubmits it, adapts it — permanently, forever, as long as it lives.
None of this is secret. It’s just invisible from the outside, because none of it happens in the first afternoon. It happens in year one, year two, year three. Internally, we have a name for all of it: the Wall. Everything that stands between a demo that works and an app people actually carry in their pockets — store review, OS updates, notifications, payments, privacy rules, performance once the data is real, and the plain passage of time. None of those bricks is optional, and not one of them appears in the demo. And hitting the Wall always feels the same: you had something that looked 90% done, and you discover the missing 10% was actually the bulk of the work.
The AI generators haven’t made any of this disappear. They’ve done something stranger: they’ve multiplied, by an order of magnitude, the number of people who reach the Wall. There have never been so many beautiful prototypes standing in front of it.
The part that surprised even me
If the story stopped there, it would be a comfortable one for someone in my position: AI makes prototypes, we do the serious part, everyone go home. But something happened this year that I didn’t see coming, and it’s changed how I think about the whole question.
The AI tools started connecting to the platforms they were supposed to replace.
There’s an open standard now — MCP, for those who follow this — that lets an AI agent operate real software directly. Our engineers plugged our platform into it, and I’ve watched what that looks like in practice: you can sit in an AI assistant and manage an actual, live, published app. Update the catalogue. Schedule a push notification. The agent doesn’t rebuild the app from scratch; it works with the one that already exists, the one that’s already past the Wall.
This didn’t come out of nowhere, by the way. Our CTO recently wrote about what happened when they let an AI write code inside our platform — his conclusion, roughly, was that generating the code was the easy part. Everything I’m describing here is the same lesson, seen from the marketing seat instead of the engineering one.
I’m careful with predictions — I’ve been wrong before, see above — but this reversal seems solid to me: the better AI gets at doing things, the more it needs real, running, maintained software to do things with. A prototype is a dead end for an agent. There’s nothing to operate.
The threat didn’t vanish. It turned into a door, and it walked through it.
Where I’ve landed
So, will AI kill no-code app builders? Here’s my honest scorecard, from inside.
What AI has already killed: the cost of trying. The blank page. The three months and the budget line you used to need before you could even look at your idea. Good riddance, honestly — I’ve spent years watching people give up at that stage, and I won’t miss it.
What it hasn’t touched: everything that makes an app exist over time rather than merely appear on a screen. The stores, the updates, the payments, the year two. That work has moved around, it’s better tooled than it was, but somebody still owns it — and pretending otherwise is how you end up with a folder of dead prototypes.
If you’re building something right now, my only real advice is this: enjoy the magic hour, it’s real. Then, before you tell anyone your app is almost done, ask yourself who is going to be maintaining it eighteen months from now, and on what. However you answer that question, you’ll have skipped the most expensive mistake of this whole era — believing that because the beginning got easy, the rest did too.
The industry won’t tell you this because both sides have cleaner stories to sell. The truth is messier: nobody died. The hard part just became visible.
Muriel Santoni — Marketing & GEO @ GoodBarber. Storyteller. I spend my days making a no-code app builder easy to find and hard to forget. I write about no-code, AI, and what happens when a non-dev starts to build.