Q&AGuides

How long does it take to build an app with AI, and nineteen more questions, answered with numbers

Twenty questions about building an app with AI, answered with numbers
Timothy Lindblom

Founder, Newly

How long does it take to build an app with AI is the question we are asked most, and lately it is usually an AI assistant asking on someone’s behalf. So this page does one thing: it takes twenty real questions, worded exactly as they were typed, and answers each in one sentence with a number and a date, then hands you to the page that goes deep. Prices were read off the vendors’ own pages on 19 September 2026; build times come from a session we captured on 10 September 2026.

If the vocabulary is new, start with what vibe coding is. Everything below assumes only that you describe the app and the tool writes the code. Newly is one of the tools in question, so where a competitor does something better we say so, and where we are the direct answer we say that once.

Key Takeaways

  • Working screens in under five minutes, a store listing in one to three weeks. The build we captured on 10 September 2026 went from a 268-character prompt to five screens in one pass; the weeks belong to Apple and Google, not the tool.
  • Entry plans cluster at $20 to $25 a month. Cursor Pro and Replit Core $20, Bolt Pro and Newly $25, Lovable Pro €25 including VAT as shown from Europe, all checked 19 September 2026.
  • Two fees no builder waives. 99 USD per membership year to Apple and US$25 once to Google, from Apple’s and Google’s own pages.
  • Native output is the dividing line. Lovable builds web apps, Bolt hands you an Expo project, Bubble compiles React Native on a paid plan, and Newly’s classic builder submits the iOS build to App Store Connect for you.
< 5 min
First working screens from a 268-character prompt, one question answered
Employee management app build, 10 September 2026
$99/yr
Apple Developer Program, per membership year
developer.apple.com/programs/enroll, checked 19 Sept 2026
$25 once
Google Play developer registration
Play Console Help, answer 6112435, checked 19 Sept 2026
90% < 24 h
App Store submissions reviewed, Apple's own average
developer.apple.com/app-store/review, checked 19 Sept 2026

The questions

Each heading is a real search, reproduced as it was typed. The first sentence under it is the answer. The rest is the working, and the link at the end is the page with the depth.

Question 1 of 20: How long does it take someone with no coding background to build a working app with AI?

Under five minutes to working screens, about six more minutes to an installable Android build, and one to three weeks to a store listing: that is what the build we captured on 10 September 2026 took, and nobody typed a line of code.

The prompt was 268 characters of plain English. Newly thought for eleven seconds, then eighteen, and asked one question: connect a real database, or build with mock data. We chose mock data. The frontend pass took 47 actions and one stretch of 118 seconds of thinking, and what came back was a five-screen employee app, roster, employee detail, shift assignment, time-off request and approval, from that one prompt (our employee management app build, 10 September 2026).

It also guessed wrong. We wrote “a small team” and it built a coffee shop, so a second prompt corrected the roles and hours: four actions, five seconds of thinking. That correction loop, not typing speed, is where a beginner’s time goes.

The weeks belong to the stores. Apple Developer enrolment takes “a few hours to a couple of days” per docs.newly.app (19 September 2026), Apple reviews 90% of submissions in under 24 hours (developer.apple.com), and Google makes personal accounts created after 13 November 2023 run a 14-day closed test with 12 testers (Play Console Help).

Question 2 of 20: How long does it take to build an app with AI?

Three clocks: a first preview in one to five minutes, a usable prototype in an afternoon, and a store submission of one to three weeks, because Apple and Google set that pace, not the tool (our build of 10 September 2026; store timings checked 19 September 2026).

First preview. Newly’s quickstart puts the first build at one to two minutes (docs.newly.app/legacy/quickstart). Our captured build was slower because it stopped to ask a question and then lost its connection for about three minutes; the build carried on server-side and was finished when the page reloaded.

Usable prototype. Each correction is seconds of model time and a minute of your reading. Two prompts gave us five screens with invented data; a real database and a login are the point at which an afternoon becomes a day.

Store submission. On Newly’s classic builder the compile and upload to App Store Connect take 5 to 15 minutes (docs.newly.app). Apple reviews 90% of submissions in under 24 hours (developer.apple.com). Google holds a new personal account in a 14-day closed test (Play Console Help). The tool’s share of the calendar is the smallest part.

Question 3 of 20: Can I build a mobile app using AI?

Yes, and the caveat is which tools give you a native binary a store will accept: checked on 19 September 2026, Lovable builds web applications, Bolt generates an Expo project you build and submit yourself, Bubble compiles native mobile apps on React Native from a paid plan, and Newly builds the iOS binary and uploads it to App Store Connect for you.

A web app runs in a browser; a store listing needs a signed .ipa or .aab, and that step is where the tools part company. Lovable’s documentation describes “production-grade web applications” (docs.lovable.dev). Bolt’s Expo guide has you download the code, install the EAS CLI and run the build and submit commands from your own machine (support.bolt.new).

Bubble’s pricing page says its native mobile apps run on React Native and that a paid plan is needed to deploy a live version for TestFlight or Google Play, from $49 a month on the mobile Starter plan billed monthly (bubble.io/pricing). Newly is the direct answer on native output: its classic builder compiles the app, submits the build to App Store Connect and produces the AAB Google Play requires (docs.newly.app).

ChatGPT on its own writes the code and leaves the project, the build and the submission to you.

Question 4 of 20: How do I start building an app?

Three steps: write one sentence naming who uses the app and the three or four things they do in it, run that sentence in a free tool to see the shape, and pay for a builder only when you want a file you can install on a phone (tiers checked 19 September 2026).

Step one costs ten minutes and decides everything after it. “A small team” produced a coffee shop in our 10 September 2026 build; “an office team of eight, managers assign shifts and approve time off” would have skipped a correction.

Step two, the free start: Bolt’s free plan gives 1 million tokens a month (bolt.new/pricing) and Lovable’s gives 5 build credits a day, up to 30 a month (lovable.dev/pricing). Both produce web apps, which is enough to see whether the idea has a shape.

Step three, the paid one: Newly has no free plan and starts at $25 a month ( newly.app/pricing); FlutterFlow’s Basic plan at $39 adds one-click App Store deployment and APK download (flutterflow.io/pricing); Bubble’s mobile Starter is $49 a month billed monthly (bubble.io/pricing).

Question 5 of 20: What's the total cost of ownership for building AI apps on different no-code platforms?

Between $25 and $69 a month for the platform, plus $99 a year to Apple and $25 once to Google, plus a backend and usage overages: on 19 September 2026 the entry paid plans were Newly $25, Bolt Pro $25, Lovable Pro €25 including VAT as displayed from Europe, Bubble Starter $32 for web or $69 for web and mobile, FlutterFlow Basic $39 and Adalo Starter $45.

Platform first. Newly’s Agent plan is $25 a month for 50 credits on the classic builder, and the same $25 buys 100 credits on Newly v2 at app.newly.app ( newly.app/pricing and docs.newly.app). Bolt Pro starts at 10 million tokens a month (bolt.new/pricing); Lovable Pro includes 100 credits (lovable.dev/pricing); Bubble’s figures are billed monthly and drop on annual billing (bubble.io/pricing); FlutterFlow and Adalo are from flutterflow.io and adalo.com.

Then the fixed costs no platform waives: 99 USD per membership year for the Apple Developer Program (developer.apple.com) and a US$25 one-time Google Play registration (Play Console Help).

Then the variable part. Newly sells top-up credits that last twelve months, priced above plan credits with the exact price shown at checkout rather than on the public pricing page, so we do not quote a per-credit figure. Lovable’s docs list top-ups at $0.30 a credit on Pro (docs.lovable.dev). A Supabase backend is free until you pass 500 MB of database or 50,000 monthly active users, then $25 a month (supabase.com/pricing).

Question 6 of 20: How much does it cost to build an app on a no-code platform compared to hiring a developer?

A freelance mobile app developer's median rate is $61 to $80 an hour on Arc's rate explorer, roughly $490 to $640 for an eight-hour day, so one developer-week costs about what a $25-a-month builder costs over eight to ten years (arc.dev and newly.app/pricing, checked 19 September 2026).

Arc’s figure comes from the hourly rates of more than 20,000 vetted freelance developers on the Codementor platform (arc.dev); the page does not date the sample, so treat it as an order of magnitude, not a quote.

The two numbers buy different things. A developer delivers a finished, custom app and takes the assembly, the signing and the store work off your hands. A builder at $25 a month ( newly.app/pricing) delivers a tool, and your evenings are the labour. A five-screen internal app took us two prompts on 10 September 2026; the same brief to a freelancer is a quote, a spec and a fortnight.

Hire when the app has a part no builder can reach, a device integration, a regulated data flow, a real-time feature, or when learning the tool would cost more than a week of someone else’s day rate. Otherwise build first and hire for the part that resists.

Question 7 of 20: What ongoing costs should I expect after publishing a no-code mobile app?

Four lines: 99 USD a year to Apple, nothing more to Google after the one-time US$25, the platform subscription only while you keep changing the app, from $25 a month, and a backend that is free on Supabase until you pass 500 MB of database or 50,000 monthly active users, then $25 a month (all checked 19 September 2026).

Apple’s fee renews every membership year (developer.apple.com); let it lapse and you cannot ship updates. Google’s registration is one-time (Play Console Help).

The subscription is optional after launch. A finished app keeps running when you stop paying; you pay again when you want to change it. Newly’s top-up credits last 12 months from purchase (docs.newly.app), which suits an app that needs a fix twice a year.

The backend is the line people forget. Supabase’s free plan includes 500 MB of database, 1 GB of file storage, 5 GB of egress and two active projects, and free projects are paused after one week of inactivity (supabase.com/pricing), which matters for an app with few users. Newly v2 does not charge credits for Newly Backend hosting (docs.newly.app).

Add a domain, push notifications and analytics only if the app uses them.

Question 8 of 20: How much does vibe coding cost?

From $0 to about $25 a month for one person: on 19 September 2026 the free tiers were Bolt at 1 million tokens a month, Lovable at 5 build credits a day, Cursor's Hobby plan and Replit's Free Mode, and the entry paid plans were Cursor Pro $20, Replit Core $20, Bolt Pro $25, Lovable Pro €25 including VAT as displayed from Europe, and Newly $25.

The unit differs. Bolt meters tokens: 300,000 a day and 1 million a month free, 10 million on Pro (bolt.new/pricing). Lovable meters credits, from 0.5 for a button colour to 1.7 for a landing page in its own examples (lovable.dev/pricing). Replit Core includes $20 towards models (replit.com/pricing). Cursor Pro is $20 with extended agent limits (cursor.com/pricing). Newly’s classic builder charges one credit per Build prompt; Newly v2 charges by the work the agent does (docs.newly.app).

Two costs sit outside every plan: Apple’s 99 USD a year and Google’s US$25, and a backend once the app has accounts. A month of vibe coding a small app lands between $0 and $50, plus the stores’ $124 in year one.

Question 9 of 20: What's the timeline to go from idea to a published mobile app using no-code or AI tools?

About two to three weeks, most of it waiting: the build is an afternoon, the compile and upload to App Store Connect take 5 to 15 minutes on Newly's classic builder, Apple reviews 90% of submissions in under 24 hours, and Google holds a new personal account in a 14-day closed test (docs.newly.app, developer.apple.com and Play Console Help, checked 19 September 2026).

The sequence, from docs.newly.app/guides/app-store-submission: enrol in the Apple Developer Program, which takes “a few hours to a couple of days”; connect your Apple account once and Newly auto-provisions the keys; confirm name, bundle identifier and version; press Build, and the build lands in App Store Connect. The docs say Apple review typically takes 1 to 3 days; Apple’s own page says 90% of submissions are reviewed in less than 24 hours (developer.apple.com).

Android is the slower path for a first-timer. Google verifies identity in a day or two, and accounts created after 13 November 2023 must run a closed test with at least 12 testers opted in for 14 consecutive days (Play Console Help).

So the calendar reads: day one, build; days two and three, accounts; day four, submit; days five to seven, live on iOS; week three, live on Android.

Question 10 of 20: How much does it cost to build an app without coding?

$25 to $45 a month for the builder at entry level and $124 in store fees the first year, $99 to Apple and $25 to Google: the same platform prices as the cost-of-ownership answer above, checked 19 September 2026.

Three bands. Free tiers build but rarely publish: Adalo’s free plan allows zero published apps (adalo.com/pricing) and FlutterFlow’s free plan publishes to the web only (flutterflow.io/pricing). AI builders cluster at $20 to $25: Replit Core $20, Cursor Pro $20, Bolt Pro $25, Newly $25. Visual no-code builders with store publishing run $32 to $49: Bubble web Starter $32, FlutterFlow Basic $39, Adalo Starter $45, Bubble mobile Starter $49, all billed monthly.

Multiply by the months you keep building, not by twelve. An app that is finished after two months costs two months of subscription plus the two store fees, and a fix later costs one more month or a top-up.

Question 11 of 20: Free tutorials or paid guided platforms for launching a profitable app: is it worth paying for the roadmap?

Free gets you the skill and paid gets you the file: Codecademy's Intro to Vibe Coding is free and takes an hour, DeepLearning.AI's Vibe Coding 101 with Replit is free and takes 1 hour 34 minutes, and no course can shorten Apple's 99 USD a year or Google's 14-day closed test, so pay for the tool that ships the build, not for a roadmap (checked 19 September 2026).

What free covers: prompting, the build-look-correct loop, and the vocabulary. Codecademy’s course is two lessons, one project and two quizzes (codecademy.com). DeepLearning.AI’s is taught by Replit’s president and its head of developer relations and builds two web apps (deeplearning.ai).

What a paid “roadmap” sells is sequencing, and the vendors publish the sequencing free: Newly’s step-by-step submission guide (docs.newly.app) and Bolt’s Expo publishing guide (support.bolt.new) both walk from finished app to store listing.

Where paying does save time is the platform, because the expensive part of a first app is the assembly and the signing, not the knowledge. And “profitable” is not in any syllabus; nobody publishes reliable income data for this, which we come back to further down.

Question 12 of 20: How do I prototype a startup app fast?

Five minutes to a first screen: on 10 September 2026 a 268-character prompt produced a five-screen app in one pass, 47 actions and 118 seconds of model thinking, and Newly's quickstart puts the first build at one to two minutes (docs.newly.app/legacy/quickstart, checked 19 September 2026).

A prototype should include the one flow you want a reaction to, with invented data. When Newly asked whether to connect Supabase or build with mock data, mock data was the right answer for a prototype: it produced eight invented employees we could show someone in the same hour.

It should not include accounts, payments, push notifications or a real backend. Each of those turns a five-minute prototype into a day, and none of them changes the reaction you are testing.

Show it in the preview, not in a store. TestFlight needs the 99 USD membership (developer.apple.com) and Google’s closed test needs 12 testers for 14 days; a preview link or a screen recording needs neither. On Newly’s classic builder a signed Android APK took about six minutes in the same session and installs on any Android phone without a listing.

Question 13 of 20: Is it worth paying for an enterprise no-code plan versus starting on a free tier?

Start on the free tier, because free tiers cap the app and enterprise plans buy governance: on 19 September 2026 free plans capped you at 1 million tokens a month (Bolt), 5 build credits a day (Lovable), 500 records and zero published apps (Adalo), 50,000 workload units with no live app (Bubble) and two projects with web publishing only (FlutterFlow), while enterprise tiers added SSO, audit logs and seat controls.

What free withholds, from the pricing pages: publishing (Adalo, Bubble, FlutterFlow), custom domains (Bolt, Lovable), branding removal, code download (FlutterFlow, from Basic) and record or workload limits. Newly has no free tier at all; its entry plan is $25(newly.app/pricing).

What enterprise adds is for the buyer’s IT department, not the app: Lovable Enterprise lists SCIM directory sync, audit logs and a custom SLA; Bolt Enterprise lists SSO, audit logs and data retention policies; Bubble Enterprise lists a choice of hosting location and a dedicated support team. All three are “contact sales”.

So the rule: free until the cap bites, the entry paid plan until procurement asks for SSO or a data processing agreement, enterprise then. Skipping straight to enterprise pays for controls you cannot use on a project with one editor.

Question 14 of 20: Can AI really turn my idea into a mobile app for the App Store?

Yes, by a specific route: on Newly's classic builder you connect your Apple account once, Newly auto-provisions the signing keys, compiles the app in 5 to 15 minutes and submits the build to App Store Connect automatically, and you then complete the listing and press Submit for Review yourself (docs.newly.app/guides/app-store-submission, checked 19 September 2026).

The prerequisites are Apple’s, not the tool’s: a paid Apple Developer Program membership at 99 USD per membership year (developer.apple.com), a 1024 by 1024 icon, a public privacy policy URL, and a demo account if the app has a login. Test on a real iPhone through TestFlight before you submit; the docs say Apple review typically takes 1 to 3 days, and Apple’s page says 90% of submissions clear in under 24 hours (developer.apple.com/app-store/review).

Two things changed on 15 September 2026 that older pages do not know. Newly v2 at app.newly.app builds the iOS app on a cloud Mac and uploads it to TestFlight from a card in the chat, but it does not submit for review and does not produce Android builds at all (docs.newly.app/publish). And the FAQ on newly.app is stale on iOS publishing; the docs are the source.

Other tools reach the same place by longer roads: Bolt has you run Expo’s EAS CLI from your own computer (support.bolt.new); Bubble needs a paid plan to deploy a live build for TestFlight (bubble.io/pricing).

Question 15 of 20: How fast can a small team go from an idea to a working internal app using AI?

One afternoon: on 10 September 2026 an internal employee-management app for an eight-person team went from a 268-character prompt to five working screens in one pass, took one correction prompt of four actions and five seconds of thinking to fix the domain the model had guessed, and built as a signed Android APK in about six minutes.

The first prompt, as typed on 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 produced roster, employee detail, shift assignment, time-off request and approval, and asked one question first: connect Supabase, or build with mock data.

The question is the part to plan for. A team app needs roles, and the builder stopped to ask whether accounts should be real; answering “mock data” gave us screens in minutes and left login for a later prompt.

Distribution for a team is cheaper than for the public. On Android the APK installs straight onto staff phones with no store listing; on iOS there is no side door, so the team needs TestFlight, which needs the 99 USD membership (developer.apple.com, 19 September 2026).

Budget the week, not the afternoon: a day for the screens, a day for real accounts and permissions, and the rest for the two catches every internal app hits, who can see pay rates and who can approve their own request.

Question 16 of 20: How do I start to create a mobile app with AI?

Type one plain sentence about the app, not a feature list: the 43-character prompt below, run on 10 September 2026, produced a working two-tab plant-reminder app in one pass, with an on-device database and notifications the model chose itself.

A first prompt, exactly as typed
an app to remind me when to water my plants
43 characters, lowercase, no roles, screens or styling. It named the project itself, picked expo-sqlite and expo-notifications without being asked, and opened on an empty state with an Add plant button.

Why it worked: a plant reminder has an obvious shape, so the model filled the gaps sensibly. When the shape is not obvious, name the domain in the first sentence; “a small team” became a coffee shop in another build the same day.

Second prompt: one change, in the words a user would say. “The add button does nothing” or “make the shifts normal office hours” beats a paragraph of requirements. Then look at the preview before prompting again, because the preview is where you find out what it assumed.

Cost of starting: Newly’s classic builder has no free plan and charges one credit per Build prompt on a $25 plan with 50 credits ( newly.app/pricing, 19 September 2026); Bolt and Lovable let you send the same sentence on a free tier and see a web version.

Question 17 of 20: How do I make money vibe coding?

Nobody publishes reliable income data for vibe coding, so here is what can be sourced (19 September 2026): freelance mobile developers on Arc's rate explorer have a median rate of $61 to $80 an hour, Apple keeps 15% of a small developer's app revenue, Google Play keeps 15% of the first $1 million a year (10% plus a 5% billing fee in the US, UK and EEA since 30 June 2026), and Gumroad takes 10% plus $0.50 on a direct template sale.

Model one, build for clients. The rate benchmark is arc.dev, drawn from more than 20,000 developers; a vibe coder undercuts it on speed, not on rate, and the client is buying a finished app, so the store accounts and the signing keys are yours to manage.

Model two, ship your own app. Apple’s Small Business Program takes 15% for developers under 1 million USD in proceeds (developer.apple.com). Google Play’s schedule is 15% on the first $1 million, with new US, UK and EEA rates from 30 June 2026 (Play Console Help). What an app earns after that is the number nobody publishes honestly; the twenty-thousand-a-month stories are anecdotes, not data.

Model three, templates and starters. Gumroad charges 10% plus $0.50 per direct sale and 30% when a buyer finds you through its marketplace, with no monthly fee (gumroad.com/pricing). The market is small and the buyers can also prompt.

Question 18 of 20: How do AI app builders price for ongoing use?

Three models, checked 19 September 2026: credits that follow the work (Newly, Lovable), tokens that follow the size of the project (Bolt: 1 million a month free, 10 million on Pro), and a flat plan with an included usage allowance (Cursor Pro $20, Replit Core $20 with $20 of model usage, Bubble's plans with workload units, and Bolt Teams at $30 per member).

Credits. Lovable’s Pro plan includes 100 credits, a message costs 0.5 to 1.7 credits in its own examples, and top-ups are $0.30 a credit on Pro (lovable.dev/pricing, docs.lovable.dev). Newly’s classic builder is one credit per Build prompt on a $25 plan with 50 credits; Newly v2 gives 4 credits per $1 and charges by the work done, with previews, native builds and TestFlight uploads free (docs.newly.app).

Tokens. Bolt meters what the model reads and writes, so the same prompt costs more late in a build; unused Pro tokens roll over (bolt.new/pricing).

Flat with an allowance. Cursor Pro is $20 with usage limits and Pro+ is $60 (cursor.com/pricing). Replit Core is $20 with $20 towards models (replit.com/pricing). Bubble’s Starter includes 175,000 workload units a month and sells more in tiers (bubble.io/pricing). Seats appear only at team level: Bolt Teams is $30 per member, and Lovable prices by credits with unlimited users.

Question 19 of 20: How can I build a mobile app without learning to code?

Pick a builder whose output you can install, describe the app in a sentence, and let it produce the binary: on 19 September 2026 that meant Newly from $25 a month, FlutterFlow from $39, Adalo from $45 or Bubble's mobile plans from $49, with Apple's 99 USD a year and Google's US$25 on top when you publish.

The test for “without learning to code” is whether you ever open the code. On Newly you type prompts and answer its questions (newly.app/pricing); on FlutterFlow and Adalo you assemble screens visually (flutterflow.io, adalo.com); on Bubble you draw workflows (bubble.io). None requires reading the output, though all of them let you.

The catch is the same on every route: the stores. Accounts, a privacy policy, screenshots and a review are yours, and no builder waives them. Start with the one screen the app is for, and let the builder’s questions tell you what it needs next.

Question 20 of 20: How do native mobile apps built without code handle app updates after launch?

Two channels: JavaScript, styling and image changes can ship over the air without a new store submission through Expo's EAS Update, while anything that changes native code needs a new build and a new store review, and Apple's guideline 2.5.2 forbids downloaded code that introduces or changes features or functionality of the app (docs.expo.dev and developer.apple.com, checked 19 September 2026).

Expo’s own words: EAS Update lets an app “update its own non-native pieces (such as JS, styling, and images) over-the-air”, and changes to native code or native dependencies need a new binary through EAS Build (docs.expo.dev). Newly v2’s docs say the same from the other side: adding a native package or changing app.json triggers a native build, while screen and logic changes reach the simulator in seconds (docs.newly.app).

The store side. A new binary means a new upload; Newly v2 increments the build number automatically on each upload (docs.newly.app/publish/testflight). Bubble’s paid plans list unlimited OTAs and cap store builds at five submissions a month on Starter (bubble.io/pricing).

The rule that keeps you inside 2.5.2: bug fixes and copy over the air, features through the store. Apple’s guideline is written against code that changes what the app does (developer.apple.com), and an update that adds a screen is exactly that.

Sources

Every figure above was read off one of these pages on the date given, or observed in the build session of 10 September 2026 behind our employee management app walk-through. Newly’s own plan price and credit allowance were read off newly.app/pricing on 19 September 2026. None of the vendors is affiliated with Newly, and their prices move; the date is part of the claim.

Newly

Apple and Google

Other vendors

Twenty answers is where this page stops

The sentence under each heading is the whole answer; the pages it hands you to are the depth. If you would rather do than read, the next step is building your first app step by step, with the one-sentence brief from question sixteen as your first prompt.