Build an AI app or website business
Find a problem people already pay to fix. Prove they want your fix. Then let AI write the code.
Where this goes
Your own product, and users who come back because it fixes their problem.
A user emails to say your app saved them an hour this week, and asks for one more feature. You built that, from an idea to a live product people pay for, and every line of code sits in your own account. Each morning you check who came back, read their notes, and ship one small fix. You know your costs, your users and your next move. And you've proven to yourself you can find a real problem and build the answer.
First win. Day 1: you pick a problem people already pay to fix and write your one-line promise.
What it takes. About 5 to 20 hours a week for 3 months, with money and hour caps you stick to.
Money. Most apps earn little: the middle half of subscription apps make $16–$429 a month, and only 17% of new apps reach $1,000 a month within two years (RevenueCat, Mar 2026). These are ranges, not promises.
- Day 1: pick the problem, not the app
You’ll have 10 problems scored, one picked, a one-line promise, money and hour caps set, and your builder, GitHub and Stripe accounts open.
- Week 1: prove people want it before building much
You’ll have 10+ buyer conversations, a landing page with a waitlist or pre-order link, and only the core screen built.
- Month 1: ship, charge, secure
You’ll have sign-in, a server-side paywall, the security gate passed, policies live, a launch, and your cost per user measured.
- Month 3: keep, fix or kill
You’ll have retention read by signup month, one channel doubled, a price test, and a clear keep-or-kill decision.
Paths
Your plan picks the path that fits your answers. You can switch any time.
- Build your own AI app
You want a product where AI gives a visible result in seconds, like a photo in and an answer out. Build it as a web app first; go native iOS only after people pay.
- Build your own website business
You want a site that sells one thing from a landing page and a payment link, with no app store. Best fit with under $100 to start.
Day 1: pick the problem, not the app
5 tasksGoal. One problem picked, a one-line promise, caps set, and your accounts open.
Score your problems and pick one
Write your one-line promise
Set a money cap and an hours cap
Open your accounts: builder, GitHub, Stripe, analytics
Week 1: prove people want it before building much
8 tasksGoal. 10+ buyer conversations, a landing page with a waitlist or pre-order link, and the core screen built. Milestone: 30+ signups or 3 pre-sales (Top Set's target, not a benchmark).
Find where your buyers already gather
Book 10 buyer conversations
Run 3 buyer conversations
Write the spec and the cost math
Build your landing page with a waitlist or pre-order link
Post where your buyers gather
Build only the core screen
Week-1 check: signups or pre-sales
Month 1: ship, charge, secure
9 tasksGoal. Sign-in and a server-side paywall, the security gate passed, policies live, launched, first paying user, and cost per user measured.
Add sign-in and payments in test mode
Pass the security gate before real payments
Write your privacy policy, terms and AI notice
Deploy to production
Switch payments to live
Launch to your waitlist and communities
Answer support with drafted replies
Measure your cost per active user
Day-30 check
Month 3: keep, fix or kill
8 tasksGoal. Retention measured, one repeatable channel, a price test, and a clear keep-or-kill decision.
Read your retention, not your signups
Find your one channel that works and double it
Test a higher price, an annual plan or a hard paywall
Add a habit loop and saved history
Keep-or-kill decision
If it's growing: prepare the App Store or Google Play version
Keep clean books and transferable accounts
Weekly spend check
Every step and prompt, free with an account
Each task above opens up: why it matters, the steps, what the AI drafts, when it’s done, what to watch out for, and a copy-paste AI prompt for nearly every step. Save your plan with your email to see them. No password, no card.
Run this playbook as your plan
Top Set turns it into one to three tasks a day, sized to the hours you have. Free, and you can see your plan before you sign up.
Build an AI app or website business
- Start-up cost
- $0 to $100
- Hours a week
- 5 to 20
- First dollar
- Days with an audience; months or never without one
- On camera
- Optional
- Phone only
- No, a laptop
- Risk
- Medium
Typical outcome. Most apps earn little. The middle half of subscription apps make $16–$429 a month; only 17% of new apps reach $1,000 a month within two years, and 4.6% reach $10,000 (RevenueCat, Mar 2026). Of 130+ startups still on builder subdomains, 7.6% reported any monthly revenue (BigIdeasDB, Sep 2026). These are ranges, not promises.
First dollar. With an audience, first payers came in days. Without one, no credible median exists, and most apps never reach meaningful revenue. New personal Google Play accounts need a 14-day closed test with 12+ testers first.
Hours. With 5 hours a week, build one tiny tool for one job and expect months. The upper end is Top Set's estimate.
- You want money this month.
- You won't talk to 10+ buyers before you build.
- You want your first app to be a 'make money' or health app. Those carry extra rules.
- You can't stick to a spending cap on AI credits.
- Your idea is a thin 'chat box' on top of a general AI chatbot, or a copy of an app that exists.
What it costs. Start on free tiers with a Stripe Payment Link. Once it's live, a lean web app runs about $45–$75 a month before AI usage, plus a domain (Top Set's estimate). An iOS app adds $99 a year. AI credits are the wildcard: people report $225+ a month in top-ups and about $3,500 on one app (self-reported).
Ranges come from public data. Updated November 8, 2026.
Episode 1 walks through this method with Say It First, an app Avery built before the series. It has no users yet, because nothing was posted. That's the lesson.
See Say It First (opens in a new tab)Tools this playbook uses
In the order the steps use them. Most have a free plan or a trial: start there.
- GitHub (opens in a new tab) Code
- Stripe (opens in a new tab) Code
- Lovable (opens in a new tab) Code
- Base44 (opens in a new tab) Code
- Bolt (opens in a new tab) Code
- Cursor (opens in a new tab) Code
- Supabase (opens in a new tab) Code
- PostHog (opens in a new tab) Code
- Vercel (opens in a new tab) Code
- RevenueCat (opens in a new tab) Code
- Replit (opens in a new tab) Code
Top Set may earn a commission. It never changes what we recommend. How partner links work
Get approved for financing
Only once you have steady sales. Never borrow to start a method you haven’t proved. Your buyer pays over time. You're paid now, minus a fee. The provider (Klarna, Affirm, Afterpay and others) collects from the buyer and takes the risk if they don't pay.
Not financial advice. Terms change often: check the provider's official page for the current terms before you apply or accept.
Red flags
- 'Build an app, make $X a month' pitches. The FTC banned Growth Cave, which sold 'AI-fueled' app businesses as 'impossible to fail' ($48.6M judgment).
- Paid code starter kits. AI made starter code nearly free; one well-known kit lost about 98% of its revenue.
- Builders that won't let you export your code.
- Anyone selling fake reviews, upvotes or downloads.
- Debug loops that burn credits with no checkpoints.
- Thin wrappers and copycats. Apple rejects them (4.2, 4.3).
What gets counted
Keep these numbers in your own notes or a sheet, and check them each week. Money in and out goes in Top Set’s money log.
- Problems listed and scored
- Buyer conversations held
- Waitlist signups
- Pre-sales
- Security gate passed
- Paying users
- Monthly recurring revenue (with costs and profit beside it)
- AI cost per paying user
- Monthly retention by signup month
- Spend vs your monthly cap
Sources
Updated November 8, 2026.
- Vercel fair use guidelines (https://vercel.com/docs/limits/fair-use-guidelines), checked Nov 8, 2026: Hobby is non-commercial only; payments or advertising a product count as commercial
- MacRumors, Aug 13, 2026 (https://www.macrumors.com/2026/08/13/app-store-fees-apple-link-outs/), news report checked Nov 8, 2026: no Apple commission on US link-outs in effect; Apple proposed up to 15% (5% Small Business Program) for the court to review
- Top Set’s own research notes for this method (not public), which name the source behind every number above.