// FAQ

Vibe coding is building software by describing what you want in plain English and letting an AI coding tool write the code — you steer, it types. The term was coined by Andrej Karpathy in early 2025. We have a full plain-English guide at appideas.ai/what-is-vibe-coding.

Start from a pain a specific group of people already pays to solve, not from a technology. The test is three parts: a narrow audience you can name, a task they currently do badly by hand or in spreadsheets, and evidence they spend money on it. Every idea in this directory is built on that test.

Talk to five people in the target audience before writing any code, and ask what they currently do about the problem — if the answer is nothing, they will not pay you either. A landing page with a price on it and a "join the waitlist" button is a weekend-sized validation test. Building the MVP is the third step, not the first.

For a v1 of most web apps, yes. Tools like Lovable, Bolt, Cursor, and Claude Code turn a written specification into working code, and the result depends far more on the quality of the spec than on your coding ability. That is why every idea here ships with a complete build prompt instead of just a description.

A complete written spec you paste into an AI coding tool: the app name, the core user flow step by step, the exact tech stack, the screens, and the pricing model. Ours run 150-250 words, which is enough for the tool to produce a working first version without guessing.

Rough guide: Lovable or Bolt if you have never coded, since they run in the browser and handle hosting. Cursor or Claude Code if you are comfortable in an editor and want more control. Every build prompt on this site works with any of them.

Each idea is labelled weekend, 1 week, or 1 month: the realistic time for one person using AI coding tools to reach a shippable v1, including the boring parts like payments and deployment. Mobile apps run longer than web apps because of app store review and device testing.

Usually close to nothing at the start. Free tiers from Cloudflare, Vercel, and Supabase comfortably cover an app with hundreds of users; a custom domain is around $10-15 a year. Costs only become real once you have enough users to be earning revenue anyway.

Yes. The major AI coding tools give you ownership of the code they generate, so you can charge for the product, sell the business, or open-source it. Check the terms of the specific tool you use, but selling what you build is the normal, intended use.

Yes. Build anything here, ship it, and keep every penny. There is no license, no attribution requirement, and no catch. If you do build one, email hello@appideas.ai and we will link to your project.

Ideas anchored to a real company reportedly earning meaningful monthly revenue with a similar business model. Each one shows the comparable company, its reported MRR, and an honest replicability rating — easy, moderate, or hard — so you know whether a solo builder can realistically compete before you start.

The models that work for solo builders are boring and reliable: monthly subscriptions for business tools, one-time purchases for consumer apps, per-event pricing for occasions like weddings, and free tools that feed a paid tier. Every idea page states which model fits and roughly what to charge.

Go where your niche already talks: the subreddit, Facebook group, Discord, or trade forum for that audience, and show up as a helpful member with a link, not an ad. Niche ideas beat generic ones here — it is far easier to reach every pool-service owner in one forum than to reach "everyone".

Generally you can validate and even take early revenue as an individual, then formalise once the money is real — but rules vary by country, so check locally. Do not let paperwork be the excuse that stops you shipping a $0-revenue prototype.