Build an App That Markets Itself (One Emergent Prompt)
Most people rebuild an app and then wonder why nobody downloads it. This prompt fixes that in the same pass. It reverse engineers a reference product, ships you a working mobile app, and forces a social share loop into the result screen so every user who finishes hands you the next one.
Step 1: find something that already works
Go to IdeaWave. It lists real startups and what they are pulling in each month, so you are not guessing at demand.
Ignore anything complicated. The ones worth copying are almost stupidly simple, one input and one result. The example in the video predicts your adult height. That is the whole product.
What you are looking for is a single-result app. One thing in, one interesting thing out. Those are the ones the share loop works on, because the result is the thing people want to post.
Copy the URL of the one you pick. That is your reference.
Step 2: open Emergent in mobile app mode
Open Emergent and start a new build.
At the top you get Web App and Mobile App. Pick Mobile App. This matters, the whole share loop below depends on it being a real app you can put in the stores.
⚡ Open Emergent and start a build
Step 3: the prompt
Copy the whole thing. Near the top you will see this line:
[ PASTE THE REFERENCE URL HERE ]
Replace that line with the URL you copied from IdeaWave, then send it. Do not paste the URL at the bottom or in a second message, it needs to sit inside the prompt where that marker is.
Two parts of this do the heavy lifting. It makes Emergent map every feature before it builds anything, so you get the whole product instead of just the headline feature. Then it spends most of its length on design, because the build is the easy part now and not looking generated is the hard part.
Reverse engineer the product at the reference URL below and rebuild it as a polished mobile app under my own brand.
REFERENCE URL:
[ PASTE THE REFERENCE URL HERE ]
STEP 1 - MAP EVERY FEATURE BEFORE YOU BUILD ANYTHING
Go through the reference properly and write me a complete feature list. Do not skip anything and do not summarise.
- Every screen it has, in the order a real user actually hits them.
- Every input it takes and every output it gives back.
- The exact calculation or logic behind the main result, as far as you can determine it.
- Onboarding, account creation, saved history, settings, notifications, export, sharing, and any secondary calculators or sub tools hiding in there.
- Which parts are free and which sit behind a paywall.
- What it does when the input is empty, extreme or invalid.
Show me this list before you build anything.
STEP 2 - REBUILD EVERY FEATURE, THEN BEAT IT
- Rebuild all of it. Do not ship a stripped down version with only the headline feature in it.
- Match the accuracy of the core result at minimum, and reach it in fewer taps than the reference does.
- Go feature by feature down your list and ask whether each one can be faster, clearer or more useful. Build that version instead of a copy.
- Tell me what the reference is obviously missing, then add it.
- Where the reference has a clumsy or slow flow, fix it rather than reproducing the flaw.
- Handle every edge case it handles, plus the ones it gets wrong.
- Do not copy their name, logo, colours, typeface, copy or screenshots. Same job, better execution, new brand.
STEP 3 - DESIGN. TREAT THIS AS THE PRIORITY, NOT A FINISHING PASS.
Most rebuilt apps fail because they look generated. Do not let this look generated.
- Dark first. One near black background, one off white for text, and exactly ONE accent colour used only for the primary action. Nothing else gets the accent.
- Banned because they read as AI built: purple to blue gradients, glassmorphism, drop shadows on everything, rounded corners on every element, stock icon sets, emoji as UI.
- One typeface only. Build hierarchy with weight and size, never by adding fonts. Big confident numbers for the result. Small uppercase labels with wide letter spacing for metadata.
- Space is the design. Generous padding, a consistent 8pt rhythm, and never crowd the edges of the screen.
- One screen does one job. The input screen asks. The result screen answers. No dashboards, no tab bars unless the app genuinely needs them.
- Motion: 200 to 300ms, ease out. The result lands with a small spring. Nothing else on screen animates.
- Every primary button sits in the bottom third of the screen where a thumb actually reaches.
- The result screen should look like a poster someone would screenshot without being asked to.
STEP 4 - THE SHARE LOOP. NOT OPTIONAL.
- Render the result as a self contained share card, designed to be screenshotted and posted.
- Auto write a caption for that card. Give the user a tone picker with at least three options such as FLEX, FUNNY and CURIOUS, and rewrite the caption in whichever they choose.
- One large SHARE MY CARD button underneath, opening the native share sheet.
- Print my app name and a short link on the card itself. Small, bottom corner, tasteful.
- The moment they come back from sharing, offer them one more thing to try so the session does not end there.
STEP 5 - PAYMENTS
- Integrate RevenueCat for subscriptions and one off purchases.
- The first result is free. Paywall the deeper breakdown, so they only ever see the price after the thing has already worked for them.
- The paywall follows every design rule above. Three tiers maximum, one visibly recommended.
STEP 6 - SHIP
- Prepare it for App Store and Google Play submission: icon, screenshots, store copy, privacy strings.
- The icon must be legible at the smallest size. One shape, one colour, no text in it.
Build it, run it, and show me the working app. Then show me your feature list, the result screen and the share card on their own, so I can check nothing was dropped and judge the design.Why the share loop is the whole trick
A normal rebuild needs you to run ads at it forever. This one does not.
The result is the marketing. Someone finishes, gets a card that says something about them, and posts it because it is about them, not about you. Your name is on the card. Whoever sees it wants their own result.
The tone picker is doing more work than it looks like. FLEX gets posted to stories. FUNNY gets sent to a group chat. CURIOUS gets posted as a question, which pulls replies. Same result, three completely different distribution routes, and the user picks the one that fits how they actually post.
That is the difference between an app you have to push and an app that moves on its own.
Step 4: money and stores
RevenueCat goes in with one click from inside Emergent. Give away the first result, put the detailed breakdown behind the paywall. People pay after they have already seen it work, not before.
Store submission: Emergent prepares the icons, screenshots and store copy. You still need an Apple Developer account and a Google Play account to actually publish, so start those in parallel, Apple review is the slow part.
Keep it legal
Rebuild the function, never the brand. Do not reuse their name, logo, colours, copy, screenshots or assets. Ideas and features are fair game, their brand and their content are not.
If the reference makes claims about health, money or anything regulated, write your own disclaimers rather than copying theirs.
Get the next one first
New prompts every week.
Free. The new drops and the tools behind them, before they hit the feed.
No spam · New issues Sunday · Unsubscribe anytime
Need it custom?
Want this built for you?
Tell me the idea and I’ll build it. An app, a tool, an automation. You don’t need to be technical.
Related guides
Guide
The $100K Founder Build Playbook (Fix Your Own Business With Emergent)
Emergent is paying the top three owners who fix their own business with an app, deadline July 21. The exact build playbook and every prompt I use, ready to paste.
Money
Clone a Profitable SaaS With One Emergent Prompt
Find a SaaS that's already making money, paste one prompt into Emergent, and it rebuilds the whole thing as your own brand, better, monetized, and built to outrank the original on Google.
Money
Rebuild Any App That Already Makes Money (Replit + Arrfounder)
Find a product with verified revenue, hand its URL to one Replit prompt, and get a better version back. Then the prompt that finds what the original is missing and builds it. All three prompts are inside.
Frequently asked questions
No. Emergent builds and runs it. You paste the prompt and the reference URL, then review what comes back.
One input, one interesting result. Check IdeaWave for products that already have demand, then take the simplest one on the list. Complicated apps do not share well.
Rebuilding a function is normal, every app has competitors. Copying a brand is not. Use your own name, logo, colours and copy, and write your own disclaimers.
Yes, but you need your own Apple Developer and Google Play accounts. Emergent gets the build and the store assets ready. Apple review takes the longest, so start that account early.
Stay in the same conversation and ask it to redesign just the card. That one screen is worth iterating on more than anything else in the app, since it is the part strangers actually see. More build prompts are in the clone and outrank guide.