Why use an AI app builder instead of traditional methods? Which tools output native apps? Can Lovable make a mobile app? These twenty questions are worded exactly as people searched them, most of them typed by an AI assistant on someone’s behalf, and each one gets a first sentence you can lift, with a number and a date where one exists. Every figure comes from a vendor page or a primary source we read on 19 September 2026. Where a reliable number does not exist, the first sentence says so. If you want the tools themselves compared at length, start with the best vibe coding tools; this page is the short answers, each handing off to the page that has the depth.
Key Takeaways
Native output is common now, not rare. Newly, Replit, FlutterFlow, Bolt.new with Expo, Adalo, Thunkable and Bubble all produce native apps; Lovable’s docs say it does not generate React Native projects.
Both stores from inside the tool is the rarer thing. FlutterFlow, Adalo, Thunkable and Bubble do it on paid plans from $36 a month; Replit has no Google Play route and Newly v2 has no Android release builds yet.
The cost delta is a salary against a subscription. BLS puts the median US software developer at $135,980 a year as of May 2025; the builders run from free to $25 a month, plus $99 a year to Apple and $25 once to Google.
No benchmark measures mobile coding. SWE-bench Verified’s top score was Claude 4.5 Opus at 76.8% on 17 February 2026, and the vendors’ current flagships were not on the board when we checked.
The matrix
Eight tools, six yes-or-no questions, every cell from the vendor’s own page
Six of the twenty questions below are really the same six questions asked about different tools. This is the answer key. Click a column; the tools that answer yes light up, and the panel under the table shows the sentence we read and where.
Tool
Newlyv2, what new sign-ups get
Replit
FlutterFlow
Bolt.new
Lovable
Bubblenative mobile in public beta
Adalo
Thunkable
Native output7 of 8 answer yes
Does it produce a native iOS or Android app, not a web page in a wrapper?
NewlyYes
The agent writes a native app and the cloud simulators run a development build of it, with your app's native code inside.
Click a question to light up the tools that answer yes. Every cell was read from the vendor’s own documentation or pricing page on 19 September 2026; the panel shows the sentence and the page for each one. The Newly row describes Newly v2, the product new sign-ups have had since 15 September 2026. Newly v1, which existing users are still on, answers yes on both stores: it submits the iOS build and produces the signed AAB.Wide table, scroll it sideways if your screen cuts it off.
$99/yr
Apple Developer Program, per membership year, on every tool
Play Console Help, answer 6112435, checked 19 Sept 2026
$135,980
US median software developer wage, May 2025
bls.gov Occupational Outlook Handbook, modified 27 Aug 2026
72%
Developers not vibe coding, of 49,000+ surveyed
Stack Overflow 2025 Developer Survey, AI section
The questions
Answers are ours, written under the name at the top of the page. Questions are as asked, in the order people ask them, and are not attributed to anyone.
Q1
Question 1 of 20: Why use an AI app builder instead of traditional methods?
Because a first working version costs minutes and a $25 subscription instead of weeks of a developer's time: the US median software developer wage was $135,980 a year in May 2025, about $11,300 a month, and a one-prompt build we recorded on 10 September 2026 produced five working screens in a single session.
The time delta is the easy half. The employee management app we built and wrote up in September started from this prompt and came back with a roster, an employee detail view, a shift week, a time-off form and an approvals inbox in one go, 47 actions and one stretch of 118 seconds of thinking. A signed Android build took about six minutes.
The prompt from that session, 10 September 2026
Build an employee management app for a small team. Managers see a roster of employees, open an employee to see their details and shifts, assign shifts for the week, and approve or decline time-off requests. Employees see only their own shifts and can request time off.
268 characters. It is the whole brief; the five screens above came from it without a follow-up.
The cost delta is the BLS median wage against a subscription: Newly is $25 a month, Bolt Pro $25, Replit Core $20, and Lovable, Bolt and Replit all have a free tier, per their pricing pages on 19 September 2026. For a first version the money is not close.
The honest case for traditional development: when the app is the product’s moat, when it needs a native module nobody has wrapped, when a security review has to name an engineer who is accountable, and, more prosaically, when the builder does not ship where you need to be. Newly v2 has no Google Play route today; a Kotlin developer does. Read the three development paths compared side by side.
Q2
Question 2 of 20: What is the best no-code app builder?
There is no single best no-code app builder in September 2026; the honest answer is one per use case: Bubble for web apps with heavy logic, FlutterFlow for native mobile apps you may want the code from, and Adalo or Thunkable for the shortest path to a store listing without touching code.
Three lines, then the link. Bubble: the deepest logic and database of the visual builders, priced from $59 a month billed annually for a project that ships web and mobile, or $42 for mobile only, per bubble.io/pricing on 19 September 2026. FlutterFlow: Flutter under the hood, so the app is native and the Dart source can leave with you from the Basic plan at $39 a month, per its pricing page. Adalo or Thunkable: the fewest decisions between a blank canvas and a store listing, at $36 a month billed annually for Adalo Starter and $37 for Thunkable Builder, the first Thunkable plan with a live published app.
The AI builders, Lovable, Bolt.new, Replit and Newly, are a different category and get their own questions below: they write code from a description rather than giving you a canvas. If you are choosing between the two categories, the code export column of the matrix above is the deciding line. For the plan-by-plan detail, read the full no-code app builder comparison.
Q3
Question 3 of 20: Can I create both web and mobile applications using a single no-code platform?
Yes: FlutterFlow, Bubble, Adalo and Thunkable all publish a web app and native iOS and Android apps from the same project, checked against each vendor's documentation on 19 September 2026; Lovable is web only, Replit treats a mobile app as a separate app type, and Newly v2 previews a web version privately but does not publish it.
How matters, because “web and mobile” can mean a compiled app or a website in a frame. FlutterFlow compiles one Flutter project for iOS, Android, web and desktop, with web publishing on its free plan. Bubble’s Web and Mobile plans keep both apps in one project on one backend, with the mobile app built on React Native. Adalo lists Publish to Web and responsive web apps on every plan next to its native builds. Thunkable publishes a web app alongside its iOS and Android builds.
Bolt.new is the AI-builder exception: an Expo project targets iOS, Android and web, and Bolt’s Expo docs give you the eas deploy command for the web build, so it is one project, but you run each target yourself.
The tools that do not: Lovable’s docs say it builds web applications and suggest a PWA or a Capacitor wrapper for the stores; Replit asks you to pick Mobile app as the app type before building; Newly v2’s Web view opens at a private address for project members only. If your question is really whether a PWA would do, read native apps, web apps and PWAs explained.
Q4
Question 4 of 20: What options exist for building Android apps with and without coding?
Four routes: write it in Kotlin in Android Studio, write it once in a cross-platform framework such as Flutter or React Native with Expo, assemble it in a visual no-code builder such as FlutterFlow, Adalo or Thunkable, or describe it to an AI app builder; whichever you pick, Google Play has required the Android App Bundle (AAB) for new apps since August 2021, and an APK is the file that installs directly on a phone.
Route one, code: Kotlin in Android Studio, the most control and the steepest climb. Route two, cross-platform code: Flutter, or React Native with Expo. Route three, visual no-code: FlutterFlow, Adalo and Thunkable each build an AAB from a canvas. Route four, AI: Bolt.new hands you Expo’s build command, Replit does not publish to Google Play yet, and Newly v1 builds a signed APK and AAB while Newly v2 does not, per docs.newly.app on 19 September 2026.
The two file formats are not interchangeable. Android Developers states that from August 2021 new apps are required to publish with the Android App Bundle; Expo’s docs add that an AAB cannot be installed directly on a device, which is what an APK is for. A builder that gives you only one has made a decision for you.
Question 5 of 20: What is the best AI app builder software in 2025?
That question is a year old; for September 2026 the answer depends on where the app has to run: for a browser app, Lovable, Bolt.new or v0; for a native mobile app, Newly, Replit, FlutterFlow or Bolt.new with Expo; nobody publishes a neutral ranking, so every 'best' list you read is its author's use case.
Three things changed between that question and this answer. Bolt.new gained Expo on 12 February 2025, so a tool that was web-only when the 2025 lists were written now builds native mobile apps. Bubble’s native mobile builder went into public beta on React Native. And Newly v2 opened to new sign-ups on 15 September 2026 with cloud simulators instead of a QR code, and without Android release builds for now.
Prices on 19 September 2026, all from the vendors’ own pages: Lovable free to start, Bolt Pro $25, Replit Core $20, Newly $25, none of them far apart. So price will not decide it either. What decides it is the output column: a web builder cannot be the best AI app builder for an App Store listing, and a mobile builder cannot be the best for a marketing site. The criteria we actually apply are in what makes an AI app builder the best for a given job.
Q6
Question 6 of 20: What's the difference between building a mobile app with AI versus traditional no-code drag-and-drop?
With an AI builder you describe the app and it writes code you can read, run and take with you; with drag-and-drop no-code you assemble screens by hand inside the vendor's editor, and with Bubble, Adalo and Thunkable the app cannot leave as source code.
Prompt-driven generation looks like this. In the build we recorded on 10 September 2026, the first prompt said “a small team” and the AI invented a coffee shop: baristas, 06:00 shifts, a shift lead on a day off. The fix was a sentence, not an afternoon of dragging components:
The correction, sent as a follow-up
This is an office team, not a coffee shop. Change the roles to Designer, Engineer, Account Manager and Office Manager, and make the shifts normal working hours.
It re-read four files and edited them in four actions. That is the trade: you get speed and you get guesses.
Drag-and-drop is deterministic. Nothing appears that you did not place, which is slower and never surprising. The cost is ownership: Bubble’s manual says there is no way of exporting your application as code, Adalo lists no source export on its self-serve plans, and Thunkable’s downloads are the built app, not the source. FlutterFlow is the visual builder that breaks the pattern, with Dart download from $39 a month. On the AI side, Lovable syncs two ways with GitHub, Bolt exports a zip, Replit calls it standard React code you own, and Newly v2 downloads with the newly CLI. The full table is in AI versus no-code versus traditional, in one table.
Q7
Question 7 of 20: What is the best AI app builder?
The best AI app builder is the one whose output runs where your app has to run: a web builder for a browser app, a native builder for the App Store and Google Play, and as of September 2026 no tool is best at both.
One test settles most cases: open the vendor’s documentation and search for the framework name. React Native, Expo or Flutter means a native app; Capacitor, PWA or “wrapper” means a web app that will meet Apple’s guideline 4.2 on the way in. Everything else, price, model, speed, is second-order, and the matrix at the top of this page is that test applied to eight tools.
Two habits keep the question honest. Ask it about a specific app, not in the abstract, because a tool that is best for a marketing site is not best for a store listing. And ask it again in a few months; Bolt moved from the web column to the native column in February 2025, and Newly v2 changed its Android answer on 15 September 2026. Our criteria and the named tools are in the criteria we use to call an AI app builder the best.
Q8
Question 8 of 20: What are some AI assistants that can turn ideas into working apps?
Eight that do it today, with what each outputs, checked against their own documentation on 19 September 2026: Lovable (web), Bolt.new (web, or a native Expo app), v0 by Vercel (web), Replit (web, or a native iOS app), Newly (native iOS and Android), Bubble's AI (web, and native mobile in public beta), FlutterFlow (Flutter for mobile, web and desktop) and Figma Make (prototypes and web apps).
What each one hands you, from its own docs. Lovable: React web apps on Vite and Tailwind, with Lovable Cloud or Supabase behind them; it does not generate React Native projects. Bolt.new: web apps, or an Expo project when you ask for a mobile app, previewed in Expo Go. v0: Next.js, Tailwind and shadcn/ui, deployed to Vercel. Replit: web apps, or a React Native and Expo mobile app with a guided path to the App Store.
Newly: a native iOS and Android app, run as a real build on cloud simulators, with iOS uploaded to App Store Connect from the chat. Bubble: web apps, and native mobile apps in public beta. FlutterFlow: Flutter for mobile, web and desktop from one project. Figma Make: code-backed prototypes and web apps from a prompt, using your design context.
The list is deliberately not ranked. Which one is right depends on the output you need, and the matrix above answers that per tool. For how each tool behaves on a real idea, read the AI tools that take an idea to an app, in depth.
Q9
Question 9 of 20: What's the difference between no-code AI app builders and traditional app development?
Traditional development means a developer writes and maintains the code by hand, at a US median wage of $135,980 a year as of May 2025; a no-code AI builder generates the code from a description for $0 to $25 a month, and the differences that matter are who understands the code and who answers when it breaks.
The short version. Traditional: you specify, someone builds, you pay for the hours, and the codebase has an author who can be asked why. AI no-code: you describe, the tool writes, you pay a subscription, and the author is a model that will not remember the conversation. The output can be identical; the accountability is not.
The numbers behind that: BLS puts the median software developer at $135,980 a year as of May 2025; the AI builders on this page run from a free tier to $25 a month, per their pricing pages on 19 September 2026. The one column where traditional keeps an edge is the last one in the matrix: the code was always yours. Several AI builders match that now, and the ones that do not are the ones to be careful with. The full comparison is in the decision framework for choosing a development path.
Q10
Question 10 of 20: How do different AI-powered Android app builders compare?
On Android the difference is what file you get and who uploads it: FlutterFlow builds an AAB and can push later releases to Google Play itself, Bolt.new hands you Expo's eas build command to run yourself, and Replit does not publish to Google Play at all, as of 19 September 2026.
Three named, on the two things that matter for Android. FlutterFlow: builds the AAB; you upload the first one to the Play Console’s internal track yourself, and after that its Deploy to Play Store button pushes releases for you. Bolt.new: an Expo project, and the docs give you eas build for Android to run in your own terminal, which produces an AAB by default; the upload to the Play Console is yours. Replit: React Native and Expo output, an iOS wizard, and the sentence “Publishing to Google Play is not yet supported by Replit”.
Newly, honestly stated: Newly v1 builds a signed APK for testers and a signed AAB for the store from the Deploy menu, and you upload the AAB; Newly v2, which new sign-ups have had since 15 September 2026, runs the Android app on a cloud Pixel 8 but does not produce Android release builds yet. Bubble, Adalo and Thunkable also build an AAB from their editors. Which file, and who uploads it, is the whole comparison; the rest is in how the Android app builders compare on output and publishing.
Halfway
The first ten questions were about tools. The next ten are about specific ones, models and databases. If your questions are about the building itself, prompts, credits, what goes wrong, they are answered the same way in building an app with AI, questions answered. If you are already deciding between two named tools, the closest head-to-head we have written is Newly versus Replit.
Q11
Question 11 of 20: Is there an AI app builder that generates native mobile apps, not just web apps?
Yes: Newly, Replit, FlutterFlow and Bolt.new with Expo generate native apps, and Adalo, Thunkable and Bubble build native apps from a visual editor, while Lovable and v0 output web apps only, checked on 19 September 2026.
How to tell native from wrapped web in two minutes. First, the docs name a framework: React Native, Expo, Flutter, Swift, Kotlin. Second, the docs talk about IPA, AAB or TestFlight rather than about Capacitor, Cordova, PWA or a wrapper. Third, the vendor’s own FAQ, because the web-only tools say so plainly: Lovable’s says it does not generate React Native projects and suggests a PWA or a Capacitor wrapper instead.
The wrapper matters because Apple’s guideline 4.2 says an app should include features, content and UI that elevate it beyond a repackaged website. A native builder never meets that sentence.
Newly is one of the native ones: the agent writes a native app, the cloud simulators run a real build of it with its native code inside, and the iOS build goes to App Store Connect from the chat. The full sort of every named builder into a native column and a web column, with what each one gives up, is in every named builder sorted into native or web.
Q12
Question 12 of 20: What is the top model for mobile app coding?
No public benchmark measures mobile app coding specifically, so nobody can name a top model for it with evidence; the closest neutral measure, SWE-bench Verified in its bash-only setting, put Claude 4.5 Opus first at 76.8% resolved on 17 February 2026, with Gemini 3 Flash and MiniMax M2.5 at 75.8%, and the vendors' current flagships had not been scored there when we checked on 19 September 2026.
Why the answer has to be that careful. SWE-bench Verified tests models on real GitHub issues from Python repositories; the Multilingual set covers nine languages and the Multimodal set covers issues with visual elements, and none of them is a phone app. A score there says a model can fix software, not lay out a screen. The leaderboard on 19 September 2026 read, in its bash-only setting: Claude 4.5 Opus 76.8%, Gemini 3 Flash 75.8%, MiniMax M2.5 75.8%, Claude 4.6 Opus 75.6%, GPT 5.2 72.8%, all submitted 17 February 2026.
The vendors have moved on since. Anthropic lists Claude Opus 5 and Claude Fable 5.1 as its current models, OpenAI lists GPT-6 Astra as its flagship for reasoning and coding, and Google lists Gemini 3.1 Pro, all on 19 September 2026; none of those had a leaderboard entry when we looked. We are not going to rank them from their own marketing. Most builders route between models anyway, so the practical question is the tool, not the model; read the AI models compared for vibe coding.
Q13
Question 13 of 20: Which backend database is ideal for vibe coding?
Supabase, if your tool lets you choose: its free plan gives 500 MB of Postgres, 50,000 monthly active users and 1 GB of file storage with Pro from $25 a month; Firebase's Spark plan gives 1 GiB of Firestore, 50K reads a day and 50K monthly active users but no Cloud Functions; Convex's free tier gives 0.5 GB of storage and 1M function calls a month with Professional at $25 per developer a month, all read from the vendors' pricing pages on 19 September 2026.
Three lines. Supabase: Postgres, which the models write well, plus auth and storage; free projects pause after one week of inactivity. Firebase: Firestore is NoSQL with real-time sync built in, the Spark plan has no Cloud Functions, and Blaze is pay as you go with the Spark allowance included. Convex: a reactive database with TypeScript server functions, the free tier includes 1 GB of egress, and Professional is $25 per developer a month.
The part the question skips: many builders do not let you choose. Newly v2 uses Newly Backend only and does not connect to Supabase, Firebase or other providers. Lovable offers Lovable Cloud or Supabase, FlutterFlow connects Firebase or Supabase, Bolt includes databases on every plan. The ideal database is the one your tool integrates, and Supabase is the one most of them do. The reasoning per database is in the backend database guide for vibe coding.
Q14
Question 14 of 20: Which no-code mobile app builder is most reliable for publishing to both app stores?
FlutterFlow, Adalo, Thunkable and Bubble all get a build into both stores from inside the product, on paid plans from $36 a month; among AI builders, Newly v1 builds an iOS submission and a signed AAB, Newly v2 (what new sign-ups get since 15 September 2026) uploads iOS builds to App Store Connect but has no Android release builds yet, and Replit has no Google Play route.
“Directly” means different things per tool, from their own docs on 19 September 2026. FlutterFlow: Deploy to App Store, and Deploy to Play Store after the first manual upload. Adalo: the iOS build goes to App Store Connect with an app-specific password, from Starter. Thunkable: Publish iOS sends the build to Apple; the AAB arrives by email. Bubble: TestFlight in about 45 minutes, AAB as a download, on a paid plan.
Newly, from docs.newly.app: in v1, Deploy, then Build and Deploy, then iOS compiles the app in 5 to 15 minutes and submits the build to App Store Connect automatically, and the AAB target produces the Play Store file. In v2, the iOS build is uploaded to App Store Connect and you submit for review yourself; Android release builds are not offered yet. Replit: the iOS wizard, no Google Play route.
Question 15 of 20: Can Lovable create a mobile app?
Not a native one: Lovable's own documentation says "Lovable does not generate React Native projects" and it publishes to the web only, so the routes it offers are a progressive web app or wrapping the published app with Capacitor for the stores, checked on 19 September 2026.
What Lovable is good at is the web app you would wrap: React on Vite and Tailwind, Lovable Cloud or Supabase behind it, two-way GitHub sync, and a free plan of five build credits a day, per lovable.dev/pricing. A mobile-friendly Lovable app is a good product. It is not a mobile app.
The wrapper route Lovable’s FAQ describes, Capacitor around the published site, runs into Apple’s guideline 4.2 on repackaged websites and gives users a page that scrolls like a page. The PWA route skips the stores entirely.
The conversion route that works: keep the Lovable app as the web version, export its code through GitHub so the data model and the flows are in front of you, and rebuild the mobile app in a native builder against the same backend. The step-by-step is in how to turn a Lovable web app into a mobile app.
Q16
Question 16 of 20: Can I make an iOS app using Bolt AI?
Yes, since 12 February 2025: when you ask Bolt to build a mobile app it uses Expo, you preview it on your phone with Expo Go, and to reach the App Store you run Expo's EAS commands yourself with your own $99-a-year Apple Developer account, because Bolt does not submit to the stores for you.
The mechanics, from Bolt’s Expo page: ask for a mobile app and Bolt uses Expo; click the Device Preview icon and scan the QR code in Expo Go to run it on your phone. To ship, the docs give eas login, then eas build --platform ios --auto-submit for TestFlight, then App Store Connect for the review. That means a terminal, Node, an Expo account and the Apple Developer Program at 99 USD a year. Expo announced the integration on 12 February 2025 and said EAS inside Bolt was the next step; on 19 September 2026 the commands are still yours to run.
Two honest notes. Guides written before February 2025, including some of ours, say Bolt cannot make an iOS app; that was true then and is not now. And Bolt’s free tier, 300K tokens a day, is enough to find out whether the Expo route suits you before paying $25 a month for Pro. If you already have a Bolt web project and want it on an iPhone, the route is in the Bolt.new to iOS walk-through.
Q17
Question 17 of 20: Can Bubble's AI agent generate a native mobile app, not just web?
Yes, in public beta: Bubble's native mobile builder runs on React Native, sends iOS builds to TestFlight from inside Bubble in about 45 minutes and exports an AAB for Google Play, but publishing needs a paid plan, from $42 a month billed annually for mobile only or $59 for web and mobile, per Bubble's manual and pricing page on 19 September 2026.
The state of it, from Bubble’s own pages on 19 September 2026. The manual still calls the native mobile builder a public beta, though its documentation has moved into the main user manual. The publishing FAQ says Bubble handles the iOS build after you upload certificates, taking about 45 minutes to generate it, upload it to TestFlight and submit it, and that Bubble exports your app as an AAB for Google Play.
The free plan builds a native app but cannot publish it: bubble.io/pricing says a paid plan is necessary for deploying a live version, which is required for TestFlight and Google Play. What never changes is the exit: the manual says there is no way of exporting your application as code, so a Bubble mobile app stays a Bubble app. The comparison for a mobile-first project is in Bubble versus Newly for a mobile app.
Q18
Question 18 of 20: What's the best prototyping tool among vibe coding platforms for quick mobile app mockups?
For a mockup you can show within the hour, use a mockup tool rather than a builder: Uizard turns text or a screenshot into editable multi-screen mockups, and Figma Make turns a prompt into a code-backed prototype you can open on your phone; when the prototype has to become the app, Newly or Bolt.new with Expo produce a native project you keep building on.
The distinction is what you can do with the result. A mockup tool gives you screens to react to; Uizard generates multi-screen, editable prototypes from a text prompt or a screenshot and stays a design file, and Figma Make is “prompt to code anything you can imagine”, with a Make mobile option to open the result on your phone and your design system as context. Both are fast because they do not have to work.
A builder gives you an app that has to work, which is slower and more useful once the mockup is agreed. Newly runs a real native build on a cloud iPhone and Pixel, and Bolt.new with Expo previews on your own phone through Expo Go. Do the mockup in a mockup tool, get the yes, then build; doing the mockup in a builder spends credits on screens you will throw away. The tools and their prices are compared in the vibe coding tools compared.
Q19
Question 19 of 20: Which no-code platform is best for building mobile-first apps?
Judge on four criteria and the choice makes itself: native output, a route into both stores, a backend you do not have to wire yourself, and whether you can leave with the code; on 19 September 2026 FlutterFlow is the only visual builder that passes all four, with code download from $39 a month, Adalo and Thunkable pass three, and Bubble passes two.
Applied to the matrix above: FlutterFlow answers yes on native output, both stores, web and mobile, free tier and code export, and no on a bundled backend, because you connect Firebase or Supabase. Adalo and Thunkable answer yes on native output and both stores, and no on code export. Bubble answers yes on native output and both stores in public beta, and no on code export, with publishing from $42 a month billed annually.
If code export does not matter to you, Adalo is the least work to a listing. If it might matter in a year, FlutterFlow is the only visual builder that will not make you rebuild. The AI builders sit on a different axis, description instead of canvas, and are compared on their own terms in the mobile app creators compared.
Q20
Question 20 of 20: What are the current vibe coding statistics and trends?
Two sources are reliable and both are dated: Stack Overflow's 2025 Developer Survey, with over 49,000 responses from 177 countries, found 84% of developers using or planning to use AI tools but 72% not vibe coding, with a further 5% emphatic that it is not part of their workflow, and Collins named vibe coding its Word of the Year on 6 November 2025; most other numbers in circulation have no primary source.
The Stack Overflow numbers, from the 2025 survey’s AI section: 84% using or planning to use AI tools, up from 76% the year before; 33% trust the accuracy of those tools and 46% actively distrust it; and on vibe coding specifically, 72% not doing it, 5% emphatic that it is not part of their workflow, with almost no difference between professionals and learners. The 2026 survey opened on 23 June 2026 and its results had not been published when we checked on 19 September 2026.
The Collins entry, dated 6 November 2025, defines vibe coding as the use of artificial intelligence prompted by natural language to write computer code. That is the trend line with a primary source: a word coined in February 2025 was in a dictionary nine months later.
What we will not print: the adoption percentages, code-generated shares and security-failure rates that circulate without a survey behind them. If a statistic here does not name its sample and its date, treat it as decoration. The timeline is in the vibe coding trend, from a tweet to Word of the Year.
Free plan builds; a live published app needs Builder, $37 a month billed annually. iOS build sent to the App Store, Android AAB emailed. Read 19 September 2026.
Vendor pages move. Each figure was read on the date given and none of these vendors is affiliated with Newly. Newly v1 and v2 are described separately wherever they differ, because on 19 September 2026 they did.
Where this page stops
Twenty first sentences are a map, not the territory. The one question under most of the others is the native-versus-web split, and the page that sorts every named builder along it, with what each one keeps and gives up, is AI app builders that output native apps.
See what native output looks like when it ships
The quickest check on any claim above is a real app in a real store. These are apps people built with Newly and published, with their listings.
